View Full Version : Optimization Statistical proof against champion [no actual impact]
EDIT: I changed the title to reflect the consensus. here's my take on the thread tl;dr:
The damage added both fighters archetype offer a small advantage.
This advantage must be present 95% of the time to be significant, which is not the case.
The randomness of the dice has a larger impact:
- you can have too many lucky rolls of 1 or 20,
- you can do 12 damage on a creature with 4 hp left.
This chaos drowns any gain from the level 3 archetype features.
END EDIT
Here's a bit of python code to simulate encounters. It runs 100k fight between 2 fighters and 4 apes and return how often each character survives, with how many hp left on average.
Fighter is level 5, scale mail + shield, dueling style rapier with Dex18 / Con16.
We run 3 times, once to raw fighter (no archetype), once for champion, once for BM (with a single SD).
Results are:
**** raw fighter
f0 98.121 46.034977222
f1 78.56 20.0855142566
m0 1.879 16.096327834 <=== apes wins the encouters almost 2% of the time
m1 1.133 12.9461606355
m2 0.315 9.18412698413
m3 0.024 6.125
percent out of 100000 tries
**** champion
f0 98.248 46.2175616806
f1 79.727 20.4124575113
m0 1.752 15.9235159817
m1 1.01 12.8118811881
m2 0.266 9.37969924812
m3 0.015 4.33333333333
percent out of 100000 tries
**** BM
f0 99.348 47.2930305592
f1 85.422 22.2417995364 <== BM #2 stays up 85% of the time, with 22 hp left on average
m0 0.652 15.7331288344
m1 0.38 12.2578947368
m2 0.1 8.75
m3 0.005 4.8
percent out of 100000 tries
Fighter 1 goes down 15-20% of the time:
raw: f1 78.56 20.0855142566
cha: f1 79.727 20.4124575113
bm: f1 85.422 22.2417995364
Even if all apes concentrate on the second BM, it will stay up 85% of the time and end the battle with 22 hp on average. That's 5% point better and 2 more hp than champion.
Improved critical has little impact on survival.
Apes win almost 2% of the time:
raw: m0 1.879 16.096327834
cha: m0 1.752 15.9235159817
bm: m0 0.652 15.7331288344
The last ape is almost 3 times as likely to win the encounter against champions.
from random import randint
import numpy as np
def fighter_rapier():
damage = randint(1, 8) + 6
extra = randint(1, 8)
return damage, extra
def fighter_rapier_sd():
damage, extra = fighter_rapier()
damage += randint(1, 8)
extra += randint(1, 8)
return damage, extra
def monster_damage():
""" ape """
return 6, 3
def try_riposte(target):
if target['sd'] > 0 and not target['riposte']:
target['riposte'] = True
target['sd'] -= 1
def encounter(team1, team2, log):
creature_list = team1 + team2
for creature in creature_list:
creature['init'] += randint(1, 20)
creature['round'] = -1
creature['riposte'] = False
# creatures will attack in initiative order
creature_list = sorted(creature_list, key=lambda o: o['init'], reverse=True)
rround = 0
while True:
rround += 1
log('round {}'.format(rround))
for creature in creature_list:
# creatures is down, skip it
if creature['hp'] <= 0:
continue
targets = team1 if creature in team2 else team2
attack_list = creature['attacks'][:] # copy, since we modify it
if creature['surge'] and rround == 1:
attack_list.extend(creature['attacks'])
if creature['riposte']:
attack_list.append(creature['sd_attack'])
for attack in attack_list:
# always target the last enemy
target = targets[-1]
attack_roll = randint(1, 20)
if attack_roll >= creature['crit']:
try_riposte(target)
if creature['sd'] > 0 and attack != creature['sd_attack']:
damage, extra = creature['sd_attack']()
creature['sd'] -= 1
else:
damage, extra = attack()
target['hp'] -= damage + extra
log('{} crits {} for {}'.format(creature['n'], target['n'], damage + extra))
elif attack_roll + creature['hit'] >= target['ac']:
try_riposte(target)
damage, extra = attack()
target['hp'] -= damage
log('{} hits {} for {}'.format(creature['n'], target['n'], damage))
else:
log('{} misses {}'.format(creature['n'], target['n']))
if target['hp'] <= 0:
log('{} down'.format(target['n']))
# target is down, remove from team1 / team2 list
target['round'] = rround
targets.pop()
if len(targets) == 0:
return creature_list
creature['riposte'] = False
def stats(repeat, char, char_count, monster, monster_count):
def log(s):
print s
if repeat > 1:
log = lambda s: 0
survival = {}
survival.update({'f{}'.format(i): [] for i in range(char_count)})
survival.update({'m{}'.format(i): [] for i in range(monster_count)})
for i in range(repeat):
team_fighter = [dict(n='f{}'.format(i), **char) for i in range(char_count)]
team_monster = [dict(n='m{}'.format(i), **monster) for i in range(monster_count)]
creature_list = encounter(team_fighter, team_monster, log)
for creature in creature_list:
if creature['round'] == -1:
survival[creature['n']].append(creature['hp'])
for k in sorted(survival):
print k, len(survival[k]) * 100. / repeat, len(survival[k]) and '{:.2f}'.format(np.average(survival[k]))
print 'percent out of {} tries'.format(repeat)
# rapier, dueling style (+2), Dex18 (+4), half-plate + shield """
fighter = dict(
attacks=[fighter_rapier, fighter_rapier], sd_attack=fighter_rapier_sd,
init=4, hp=49, ac=19, hit=6, crit=20, sd=0, surge=True)
ape = dict(
attacks=[monster_damage, monster_damage],
init=2, hp=19, ac=12, hit=5, crit=20, sd=0, surge=False)
print '**** raw fighter'
stats(10000, fighter, 2, ape, 6)
print '**** champion'
fighter['crit'] = 19 # champion crits on 19
stats(10000, fighter, 2, ape, 6)
print '**** BM 1 SD'
fighter['crit'] = 20
fighter['sd'] = 1 # BM with a single SD
stats(10000, fighter, 2, ape, 6)
print '**** BM 2 SD'
fighter['crit'] = 20
fighter['sd'] = 2 # BM with a 2 SD
stats(10000, fighter, 2, ape, 6)
Naanomi
2017-06-27, 10:20 PM
Of course, critical boosting features work best on large dice weapons, generally polearms or two-handed swords
Here are the results for 5 apes, 1-2 SD
**** raw fighter
f0 84.07 39.2859521827
f1 42.79 13.7151203552
m0 15.93 17.194601381
m1 12.02 15.4168053245
m2 6.06 12.8712871287
m3 1.77 9.23163841808
m4 0.13 4.84615384615
percent out of 10000 tries
**** champion
f0 85.12 39.3867481203
f1 44.63 14.0681156173
m0 14.88 17.2735215054
m1 11.06 15.1962025316
m2 5.59 12.8801431127
m3 1.45 8.70344827586
m4 0.06 2.66666666667
percent out of 10000 tries
**** BM
f0 93.86 42.1900703175
f1 56.83 15.9324300545
m0 6.14 16.8680781759
m1 4.26 14.6549295775
m2 1.87 12.2834224599
m3 0.37 9.13513513514
m4 0.02 8.0
percent out of 10000 tries
**** BM 2
f0 95.99 43.580060423
f1 64.08 16.481741573
m0 4.01 16.7082294264
m1 2.65 14.4528301887
m2 1.03 12.4951456311
m3 0.23 7.91304347826
m4 0.0 0
percent out of 10000 tries
Apes wins:
fighter = 16%
champion = 14%
BM/1 SD = 6%
BM/2 SD = 4%
EDIT: fixed riposte to add 1d8 damage
Of course, critical boosting features work best on large dice weapons, generally polearms or two-handed swords
Right...
Using 1d12, no shield
**** raw fighter
f0 75.98 36.3737825744
f1 31.78 11.9200755192
m0 24.02 17.392173189
m1 18.16 15.656938326
m2 9.74 12.9928131417
m3 2.8 9.93214285714
m4 0.26 6.96153846154
percent out of 10000 tries
**** champion
f0 77.09 37.2766895836
f1 34.59 12.2844752819
m0 22.91 17.5443037975
m1 17.94 15.7112597547
m2 9.7 12.8670103093
m3 2.47 9.9028340081
m4 0.25 5.72
percent out of 10000 tries
**** BM
f0 89.19 40.2694248234
f1 45.94 13.710491946
m0 10.81 16.7974098057
m1 7.42 14.3692722372
m2 3.08 12.0097402597
m3 0.75 10.2266666667
m4 0.15 9.33333333333
percent out of 10000 tries
**** BM 2
f0 92.64 41.6185233161
f1 54.1 14.2149722736
m0 7.36 16.566576087
m1 4.85 15.4103092784
m2 2.28 12.6929824561
m3 0.57 9.98245614035
m4 0.08 4.875
percent out of 10000 tries
5 apes win 23% of the time, 11% against BM
EDIT: bad values, I left dueling in there...
With defense style:
**** raw fighter
f0 73.11 35.9048009848
f1 29.76 11.7620967742
m0 26.89 17.479360357
m1 20.87 15.6641111644
m2 11.15 13.0735426009
m3 3.12 10.1794871795
m4 0.3 6.16666666667
percent out of 10000 tries
**** champion
f0 74.98 36.8811683115
f1 33.19 12.7423922868
m0 25.02 17.4440447642
m1 19.25 15.6358441558
m2 10.0 12.802
m3 2.78 9.37050359712
m4 0.31 5.8064516129
percent out of 10000 tries
**** BM
f0 87.55 39.6319817247
f1 43.87 13.7405972191
m0 12.45 16.7357429719
m1 8.56 14.4778037383
m2 3.69 12.6504065041
m3 0.85 8.97647058824
m4 0.05 5.8
percent out of 10000 tries
**** BM 2
f0 91.86 41.0774003919
f1 51.68 14.4198916409
m0 8.14 16.3783783784
m1 5.28 14.6325757576
m2 2.25 13.1066666667
m3 0.6 9.28333333333
m4 0.09 7.66666666667
percent out of 10000 tries
Theodoxus
2017-06-27, 10:29 PM
I know it's like hip and all to be as mysterious as f'n possible, but if you want us to understand wtf you're talking about, can you provide a key to what the results are showing?
What is f0, f1, m0, m1 etc.
Sorry to be a luddite grognard in your stroking of the python, but really, it's all gibberish currently.
Crgaston
2017-06-27, 10:42 PM
I know it's like hip and all to be as mysterious as f'n possible, but if you want us to understand wtf you're talking about, can you provide a key to what the results are showing?
What is f0, f1, m0, m1 etc.
Sorry to be a luddite grognard in your stroking of the python, but really, it's all gibberish currently.
Ditto. And how exactly does having an archetype apparently DECREASE the Champ's survivability?
Nifft
2017-06-27, 10:49 PM
I know it's like hip and all to be as mysterious as f'n possible, but if you want us to understand wtf you're talking about, can you provide a key to what the results are showing?
What is f0, f1, m0, m1 etc.
Sorry to be a luddite grognard in your stroking of the python, but really, it's all gibberish currently. Not him, but:
f0 = a Fighter
f1 = the other Fighter
Both Fighters are identical (i.e. of the same archetype).
m0 = an ape (because "monkey" maybe? dunno.)
m1 = other ape
m2 = penultimate ape
m3 = the final ape
Ditto. And how exactly does having an archetype apparently DECREASE the Champ's survivability?
I think he's running Monte Carlo simulations, and each simulation is a random walk through a large number of trials. Since it's random, you'd expect some small random variances.
But yeah, it seems like the number of trials may be too small, if the addition of an archetype actually decreases performance.
On average, the statistical direction of critting more should be in the Fighters' favors.
I know it's like hip and all to be as mysterious as f'n possible, but if you want us to understand wtf you're talking about, can you provide a key to what the results are showing?
What is f0, f1, m0, m1 etc.
Sorry to be a luddite grognard in your stroking of the python, but really, it's all gibberish currently.
f0, f1 are fighter 0 and fighter 1
m0...m4 are the monsters (ape in this case)
Fighter will attack m4 first and will fight until the last ape (m0) is down.
Monsters behave the same way. This is why f0 ends 98% of the battles standing while f1 goes down more often.
So, reading:
**** raw fighter
f0 98.121 46.034977222
f1 78.56 20.0855142566
When the fighters aren't using archetype features (EK without spells, BM without SD),
The last fighter will be left standing 98% of the time with 46 hp
The other fighter... 78% with 20 hp.
Apes will do about 30 damage on the other fighter when they can't down it.
So, apes will do an average of 30 damage 80% of the time, when they don't "kill" the first fighter. The other 20% of the time, it should be around 65 damage... I think {49 + (49-46) * 1/(.98 - .78) ~ 3 * 5}
Ditto. And how exactly does having an archetype apparently DECREASE the Champ's survivability?
I'm sorry. I don't see any decrease anywhere...
Crgaston
2017-06-27, 11:04 PM
I'm sorry. I don't see any decrease anywhere...
You're correct. I misinterpreted.
mephnick
2017-06-27, 11:07 PM
Oh god who cares.
The next person that makes a thread about the Champion gets pistol-whipped.
busterswd
2017-06-28, 12:26 AM
So, few things, bearing in mind I don't really know python syntax.
-Ape damage seems high. Range should be 4-9, not 6-9.
-Fighter AC is a point low; there's no reason a level 5 dex fighter shouldn't have half plate at this point. Most of the BM's power is frontloaded, while a Champion scales ever so slightly better with passives. That's going to skew your results in the BM's favor a bit.
-I don't see a limit on superiority dice for your ripostes (again, don't really read python, so could have missed it); considering your fighters have 49 HP and you have your apes swinging for 7.5 damage on average, that's a LOT of free ripostes.
Overall:
-This thread should be an object lesson in the importance of comments in your code. Unless somebody understands python AND DnD rules, there's no way of verifying what you're trying to even accomplish, much less whether your code is correct.
alchahest
2017-06-28, 12:28 AM
Oh god who cares.
The next person that makes a thread about the Champion gets pistol-whipped.
You ever try.... not clicking on threads you aren't interested in?
Foxhound438
2017-06-28, 12:56 AM
Oh god who cares.
The next person that makes a thread about the Champion gets pistol-whipped.
*nods in approval*
but jokes aside, I have one big problem and one small one...
the small problem is that this shows exactly 2 possible encounters (vs. 4 apes and vs. 5) out of -infinite-. I'm no fan of champion, particularly improved crit being the only thing they get at L3, but there may be encounters where the champion is actually better- for example, zombies. A horde of zombies will be slightly easier for champions since they get to ignore undead fortitude twice as often. Another thing might be that you're being "skirmished" against by something, the enemies have 3-4 seperate parties, the first one comes in to draw out resources, flees, short rests while the second comes in to fight you and draw out a little more, they leave, the third group comes in and you're out of gas. Out of gas, champion is going to be better than BM, even if it's not by much.
The big problem is that you're only allowing the 3rd level feature to matter and declaring the whole subclass useless. Again, I'm not a big fan of champ, but the regeneration ability is actually bonkers where it turns on. I wouldn't trust any simulation's assessment of whether or not that's better than 2-3 maneuvers in a fight though, since high level encounters can be so complex in play that a simulation of this caliber could never really be satisfactory in determining if any ability is good or not, or in what number of unlimited situations you might see all of or none of... if you catch my drift.
Finally, improved crit is a lot more effective when you have advantage- 19% crit chance vs. barely under 10% is pretty significant. This is particularly relevant for rogue/fighter builds, since crits apply to sneak attack, and the rogue can get advantage pretty consistently. (guess I had 2 small problems)
-This thread should be an object lesson in the importance of comments in your code. Unless somebody understands python AND DnD rules, there's no way of verifying what you're trying to even accomplish, much less whether your code is correct.I'll take a look at it in a few hours, when I got more time.
I would like to +1 this, however, if there happen to be any other software developers reading this.
On a sidenote: I see no reason why the results aren't outputted as integers. This would greatly improve readability, and if the descrepency is only of by 1 percentagepoint ... doesn't really seem meaningful (heck, seaing as this is a d20 based system, a 5% margin of error seems the lowest possible margin of error - if not for mathematical, for poetic reasons)
imanidiot
2017-06-28, 01:25 AM
It doesn't have to be balanced. If there weren't any "bad" choices there wouldn't be any point in having any "good" choices. Champion is fine.
JellyPooga
2017-06-28, 02:46 AM
It doesn't have to be balanced. If there weren't any "bad" choices there wouldn't be any point in having any "good" choices. Champion is fine.
That's...not the point of having choice :smallconfused:
Lombra
2017-06-28, 03:17 AM
6 fights / adventuring day with one short rest after every 2 fights, level 5 characters, 3 rounds/fight, 39 (+3 from action surge) attacks total:
The battlemaster can add averagely 12d8 of damage in an adventuring day. Critting ~2 times
An equally built champion, through crits (~4 times) and d8 weapon only adds ~2d8 during the adventuring day compared to the battlemaster.
A dedicated champion half-orc with a d12 weapon can add up to ~4d12 relatively to another half-orc battlemaster when he's lucky.
One doesn't need fancy tools or math to bring home the fact that champion isn't the best DPR fighter (I appreciate the effort that bid put in python, don't get me wrong, good job bid).
Want to add in GWM? The champion will averagely gain two more bonus attack which add 2d12+(4d12/10) + 2STR damage which is 2.4d12 of damage plus strength. Still very far from 12d8. Or not?
12d8 battlemaster vs ~6d12 plus two times strength champion. Assuming a strength mod of +3 the comparison looks like: 54 battlemaster vs 45 champion, 9 points of damage difference in favor of the battlemaster, it's not that bad actually.
Do you agree that what I wrote makes sense?
Edit: fixed some mistakes.
Edit2:fixed fixes supposed to fix mistakes but ended up being mistakes themselves.
Malifice
2017-06-28, 03:29 AM
Here's a bit of python code to simulate encounters. It runs 100k fight between 2 fighters and 4 apes and return how often each character survives, with how many hp left on average.
Fighter is level 5, scale mail + shield, dueling style rapier with Dex18 / Con16.
We run 3 times, once to raw fighter (no archetype), once for champion, once for BM (with a single SD).
Results are:
**** raw fighter
f0 98.121 46.034977222
f1 78.56 20.0855142566
m0 1.879 16.096327834 <=== apes wins the encouters almost 2% of the time
m1 1.133 12.9461606355
m2 0.315 9.18412698413
m3 0.024 6.125
percent out of 100000 tries
**** champion
f0 98.248 46.2175616806
f1 79.727 20.4124575113
m0 1.752 15.9235159817
m1 1.01 12.8118811881
m2 0.266 9.37969924812
m3 0.015 4.33333333333
percent out of 100000 tries
**** BM
f0 99.348 47.2930305592
f1 85.422 22.2417995364 <== BM #2 stays up 85% of the time, with 22 hp left on average
m0 0.652 15.7331288344
m1 0.38 12.2578947368
m2 0.1 8.75
m3 0.005 4.8
percent out of 100000 tries
Fighter 1 goes down 15-20% of the time:
raw: f1 78.56 20.0855142566
cha: f1 79.727 20.4124575113
bm: f1 85.422 22.2417995364
Even if all apes concentrate on the second BM, it will stay up 85% of the time and end the battle with 22 hp on average. That's 5% point better and 2 more hp than champion.
Improved critical has little impact on survival.
Apes win almost 2% of the time:
raw: m0 1.879 16.096327834
cha: m0 1.752 15.9235159817
bm: m0 0.652 15.7331288344
The last ape is almost 3 times as likely to win the encounter against champions.
Lol. Now run it again with a half orc champ with a greataxe.
How is one sup dice making such a difference?
Malifice
2017-06-28, 03:34 AM
6 fights / adventuring day with one short rest after every 2 fights, level 5 characters, 3 rounds/fight, 36 attacks total:
The battlemaster can add averagely 12d8 of damage in an adventuring day.
An equally built champion, through crits and d8 weapon only adds ~2d8 during the adventuring day.
A dedicated champion half-orc with a d12 weapon can add up to ~4d12 relatively to another half-orc battlemaster when he's lucky.
One doesn't need fancy tools or math to bring home the fact that champion isn't the best DPR fighter (I appreciate the effort that bid put in python, don't get me wrong, good job bid).
Want to add in GWM? The champion will averagely gain one more bonus attack which adds 1d12+(3d12/10) damage which is 1.3d12 of damage plus strength. Still very far from 12d8. Well not that far actually, it's still 12d8 battlemaster vs ~5 to 6d12 plus strength.
Do you agree that what I wrote makes sense?
Action surge adds 6 more attacks. Advantage makes crits spam more often.
Also; add a magic weapon like flame tongue to the maths. Or smite from Paladin. Or brutal crit from barbarian.
Lombra
2017-06-28, 03:41 AM
Action surge adds 6 more attacks. Advantage makes crits spam more often.
Also; add a magic weapon like flame tongue to the maths. Or smite from Paladin. Or brutal crit from barbarian.
Totally forgot about action surge, which consolidates the final damage to 6 d12+STR. And a multi-dice weapon or attack is gonna favor the champion, I just don't have the brains to make advantage comparisons on the fly tho.
Also, for the purpose of direct comparison multiclass isn't really a useful factor to take in consideration IMO.
Edit: at level 5 action surge only gives you 3 attacks/adventuring day, still confirming the 6d12+STR damage. It is also unlikely to get something like flametongue at level 5. Maybe I'll share my opinions on level 20 comparisons later, but magic items aren't really a thing to consider since they vary a lot from campaign to campaign.
edit2 for level 20 comparison:
6 encounters, 2 short rests, 3 rounds/encounter, no magic items, no multiclass, GWM, 78 attacks total (4×3×6+6).
The battlemaster adds 18d12 during the adventuring day and crits ~4 times
The champion crits ~12 times, the difference is 8 critical hits, which with a d12 weapon (both are half-orcs for simplicity's sake) equals 16d12 of extra damage. Every crit brings along another attack because of GWM, which means eight more d12 (plus crit-chance on every bonus attack, effectively adding 2d12s to the final calculation)+ 8×STR (likely 5 by 20th level).
The final comparison is:
Battlemaster: 18d12 ~117 damage
Champion: 26d12 + 40 ~209 damage
So at 20th level between two mundane fighters a champion is gonna dish out more damage than a battlemaster.
Edit: you can comfortly slap fighting styles after the considerations, champion likely gets defense and GW, increasing both survivability and damage, I don't know how to calculate the damage boost of GWFS tho.
Edit: if the character is not an half-orc the champion looses 9d12 damage from the final comparison, but keeps being better.
Did I miss something?
Err ... added 2 comments, and already something seems odd:
## returns {1d8+6, 1d8} a.k.a. the normal damage and the additional damage from a crit
def fighter_damage():
""" long sword, dueling style (+2), Str18 (+4) """
hit = randint(1, 8) + 6
crit = randint(1, 8)
return hit, crit
## returns {6, 9} a.k.a. the normal damage and the full damage from a crit
def monster_damage():
""" ape """
return 6, 9
The underlined part seems ... odd?
since, later, you don't make the distinction?
fighter = dict(damage=fighter_damage, count=2, init=4, hp=49, hit=6, crit=20, ac=18, sd=0)
ape = dict(damage=monster_damage, count=2, init=2, hp=19, hit=5, crit=20, ac=12, sd=0)
Zalabim
2017-06-28, 04:28 AM
Overall:
-This thread should be an object lesson in the importance of comments in your code. Unless somebody understands python AND DnD rules, there's no way of verifying what you're trying to even accomplish, much less whether your code is correct.
This is the big lesson here. The results are within the realm of what I expect based on the test being attempted, but I've no clue whether the test was actually programmed correctly. Like, I originally assumed ape damage 6 - 9 was referring to normal - crit damage values based on taking the averages.
Lombra
2017-06-28, 05:23 AM
I'd really love if everyone starts considering adventuring days rather than single encounters for optimization, there's no point in being useful for 1/6th of adventure. Not all encounters are fights, true, but they could be.
Hairfish
2017-06-28, 05:52 AM
That's a pretty damning conclusion. I'll certainly keep it in mind if I'm in a campaign where there are only 2 players, we both have to play the same subclass of fighter with the same gear, we top out at level 5, and all our opponents are apes.
Sirdar
2017-06-28, 05:53 AM
from random import randint
import numpy as np
def fighter_damage():
""" long sword, dueling style (+2), Str18 (+4) """
hit = randint(1, 8) + 6
crit = randint(1, 8)
return hit, crit
def monster_damage():
""" ape """
return 6, 9
def try_riposte(target):
if target['sd'] > 0 and target['riposte'] == 0:
target['riposte'] = 1
target['sd'] -= 1
def encounter(team1, team2):
creature_list = team1 + team2
for creature in creature_list:
creature['init'] += randint(1, 20)
creature['round'] = -1
creature['riposte'] = 0
creature_list = sorted(creature_list, key=lambda o: o['init'], reverse=True)
rround = 0
while True:
rround += 1
for creature in creature_list:
if creature['hp'] <= 0:
continue
targets = team1 if creature in team2 else team2
attack_count = creature['count'] + creature.get('riposte', 0)
for i in range(attack_count):
target = targets[-1]
attack_roll = randint(1, 20)
hit, crit = creature['damage']()
if i == creature['count']:
hit += randint(1, 8)
if attack_roll >= creature['crit']:
try_riposte(target)
if creature['sd'] > 0:
crit += randint(1, 8) + randint(1, 8)
creature['sd'] -= 1
target['hp'] -= hit + crit
elif attack_roll + creature['hit'] >= target['ac']:
try_riposte(target)
target['hp'] -= hit
if target['hp'] <= 0:
target['round'] = rround
targets.pop()
if len(targets) == 0:
return creature_list
creature['riposte'] = 0
def stats(repeat, char, char_count, monster, monster_count):
survival = {}
survival.update({'f{}'.format(i): [] for i in range(char_count)})
survival.update({'m{}'.format(i): [] for i in range(monster_count)})
for i in range(repeat):
team_fighter = [dict(n='f{}'.format(i), **char) for i in range(char_count)]
team_monster = [dict(n='m{}'.format(i), **monster) for i in range(monster_count)]
creature_list = encounter(team_fighter, team_monster)
for creature in creature_list:
if creature['round'] == -1:
survival[creature['n']].append(creature['hp'])
for k in sorted(survival):
print k, len(survival[k]) * 100. / repeat, len(survival[k]) and np.average(survival[k])
print 'percent out of {} tries'.format(repeat)
fighter = dict(damage=fighter_damage, count=2, init=4, hp=49, hit=6, crit=20, ac=18, sd=0)
ape = dict(damage=monster_damage, count=2, init=2, hp=19, hit=5, crit=20, ac=12, sd=0)
print '**** raw fighter'
stats(10000, fighter, 2, ape, 5)
print '**** champion'
fighter['crit'] = 19
stats(10000, fighter, 2, ape, 5)
print '**** BM'
fighter['crit'] = 20
fighter['sd'] = 1
stats(10000, fighter, 2, ape, 5)
print '**** BM 2'
fighter['crit'] = 20
fighter['sd'] = 2
stats(10000, fighter, 2, ape, 5)
@bid: Great! A small example is the first step towards a more general comparison. Add some (constructive) feedback from a forum like this one and you will soon have a really nice piece of code.
def encounter(team1, team2):
creature_list = team1 + team2
for creature in creature_list:
creature['init'] += randint(1, 20)
creature['round'] = -1
creature['riposte'] = 0
creature_list = sorted(creature_list, key=lambda o: o['init'], reverse=True)
## creature_list: the list of ppl in initiative (incl. dead creatures)
rround = 0
while True:
rround += 1
for creature in creature_list:
if creature['hp'] <= 0:
continue
## creature is current (alive) creature in init
targets = team1 if creature in team2 else team2
## targets is the list of ppl the current creature attack
attack_count = creature['count'] + creature.get('riposte', 0)
## this loops over each attack of the creature (base(2) + ripostes)
for i in range(attack_count):
target = targets[-1]
attack_roll = randint(1, 20)
hit, crit = creature['damage']()
## don't know what this is? "i" is the attack count, and a "creature['count']" is the amount of attacks of the attacker (so, ape or fighter: 2). their seccond attack gets 1d8 extra damage??
if i == creature['count']:
hit += randint(1, 8)
if attack_roll >= creature['crit']:
## if the target gets critted (or gets hit, ref code further in the line), the target uses his SD (if he has one) and, get an extra attack when it's his turn.
try_riposte(target)
## if the attacker crits, he will use an SD (if he has one) to deal extra damage
if creature['sd'] > 0:
crit += randint(1, 8) + randint(1, 8)
creature['sd'] -= 1
target['hp'] -= hit + crit
elif attack_roll + creature['hit'] >= target['ac']:
target['hp'] -= hit
## death code
if target['hp'] <= 0:
target['round'] = rround
targets.pop()
if len(targets) == 0:
return creature_list
creature['riposte'] = 0
So, to recap, the second thing that seems to be wrong:
...
## don't know what this is? "i" is the attack count, and a "creature['count']" is the amount of attacks of the attacker (so, ape or fighter: 2). their seccond attack gets 1d8 extra damage??
if i == creature['count']:
hit += randint(1, 8)
...
mgshamster
2017-06-28, 06:55 AM
Good job, bid.
Conclusion is similar to my own analysis. At all levels, BM will be doing more damage than Champ, across nearly all adventuring days. Even if it's a half-orc with a great axe and 100% advantage. It takes an extended day with about 50% more encounters or longer encounters for the champ to match the damage of the BM.
Therefore, in my opinion, the champ needs a small boost. I have always proposed a 17-20 x3 crit range, but I'm actually liking the idea of a 15-20 x2 crit range more and more. That puts them on par without requiring a half-orc with a great axe and advantage.
Lombra
2017-06-28, 07:02 AM
Good job, bid.
Conclusion is similar to my own analysis. At all levels, BM will be doing more damage than Champ, across nearly all adventuring days. Even if it's a half-orc with a great axe and 100% advantage. It takes an extended day with about 50% more encounters or longer encounters for the champ to match the damage of the BM.
Therefore, in my opinion, the champ needs a small boost. I have always proposed a 17-20 x3 crit range, but I'm actually liking the idea of a 15-20 x2 crit range more and more. That puts them on par without requiring a half-orc with a great axe and advantage.
Could you then correct my post where I reach completely different conclusions? Just for the sake of discussion, since I feel like I didn't make any huge mistakes.
DanInStreatham
2017-06-28, 07:02 AM
Hi, I think it's cool you posted the code you used. I'm not a python expert but a few specific suggestions (to add to those already):
You've used 'hit' to mean both damage and 'to hit bonus' (depending on whether it's a property of creature or in the loop). Similar issue with 'crit' which means either damage or critical threshold. This hasn't necessarily meant the code is wrong but it makes it much harder to debug.
If count is the number of attacks a creature has, why do you add 1d8 damage to hit on the second attack in the loop? I may not be understanding the purpose if that if statement.
Finally, at a high level, in this sort of stochastic simulation I think I would be more interested in the median Hp at end than the mean average, or at least interested in it.
mgshamster
2017-06-28, 07:12 AM
Could you then correct my post where I reach completely different conclusions? Just for the sake of discussion, since I feel like I didn't make any huge mistakes.
A cursory look shows that you didn't make the champ and the BM equal before doing analysis.
My own analysis starts with the extra damage capabilities of the BM, and then determining how many rounds it would take for the Champ to match it. If the number of rounds is fewer than the number of rounds in an average adventuring day, then there's a problem.
Additionally, I do not want the champ to have to optimize to reach the goals I have set.
I'll try to do a thorough analysis later, but no guarantees.
I find out the gender of my next child today; may not be on the forums that much.
Lombra
2017-06-28, 07:21 AM
A cursory look shows that you didn't make the champ and the BM equal before doing analysis.
My own analysis starts with the extra damage capabilities of the BM, and then determining how many rounds it would take for the Champ to match it. If the number of rounds is fewer than the number of rounds in an average adventuring day, then there's a problem.
Additionally, I do not want the champ to have to optimize to reach the goals I have set.
I'll try to do a thorough analysis later, but no guarantees.
I find out the gender of my next child today; may not be on the forums that much.
I assumed them equally built and just compared the only differencies between the two subclasses that could provide extra damage (superiority dies vs extended crit-range) on a fixated amount of rounds that should represent the average adventuring day. I look forward for your in-depth analysis.
Edit1: found a pretty significant mistake in the first post and edited it out. But the level 20 comparison should be accurate.
Edit2: the first sentence of edit 1 is a lie.
Congratulations and felicitations from Italy :biggrin:
Theodoxus
2017-06-28, 07:51 AM
The last fighter will be left standing 98% of the time with 46 hp
The other fighter... 78% with 20 hp.
Thank you - even labeling the table would have helped. Now at least I know what the numbers mean.
Oh god who cares.
The next person that makes a thread about the Champion gets pistol-whipped.
+1 This made me laugh.
You ever try.... not clicking on threads you aren't interested in?
+1 This made me laugh harder, as I want to say this in so many threads where someone makes a negative comment...
Good job, bid.
Conclusion is similar to my own analysis. At all levels, BM will be doing more damage than Champ, across nearly all adventuring days. Even if it's a half-orc with a great axe and 100% advantage. It takes an extended day with about 50% more encounters or longer encounters for the champ to match the damage of the BM.
Therefore, in my opinion, the champ needs a small boost. I have always proposed a 17-20 x3 crit range, but I'm actually liking the idea of a 15-20 x2 crit range more and more. That puts them on par without requiring a half-orc with a great axe and advantage.
This is an interesting idea. I've toyed with bringing the 3.P weapon crit ranges back, but haven't gotten past the thinking about it phase. I'd definitely play a Half-orc Champ fighter 3/Barbarian x with a 15-20 crit range... but is that at level 3 or 17?
mgshamster
2017-06-28, 08:08 AM
This is an interesting idea. I've toyed with bringing the 3.P weapon crit ranges back, but haven't gotten past the thinking about it phase. I'd definitely play a Half-orc Champ fighter 3/Barbarian x with a 15-20 crit range... but is that at level 3 or 17?
I think level 3 is too early. It needs to be gained over time. I don't want the extra crits to be gained on a dip.
For the past six months, I've exclusively played AL games, so I haven't had the chance to try this at home. Once my life and my player's lives all settle down, I'll do some modifications to our home game.
I'm in the process of selling one house and buying another, moving my whole family, and am about to have a third child. Meanwhile, one player has been in a series of musicals, and between practice and shows hasn't been available for game. Another player has been slammed at work doing the job of four men on a lumber yard, and he's been busy looking for a new job.
Life got really busy all at once.
With a few fixes and improvements:
- damn those half-orc barbarian apes!
- AC19
- refactor attacks
- rename damage variables
- fixed riposte crit
- action surge
- logging single encounters
- truncate decimals
Code has been updated.
Updated results:
**** raw fighter
f0 96.47 44.57
f1 70.53 17.75
m0 3.53 16.51
m1 2.41 14.45
m2 1.06 11.59
m3 0.28 8.46
m4 0.03 3.33
m5 0.0 0
percent out of 10000 tries
**** champion
f0 97.02 45.08
f1 73.07 18.22
m0 2.98 16.94
m1 2.04 14.55
m2 0.93 12.30
m3 0.22 9.14
m4 0.0 0
m5 0.0 0
percent out of 10000 tries
**** BM 1 SD
f0 99.02 46.42
f1 80.78 19.93
m0 0.98 16.57
m1 0.62 14.63
m2 0.29 13.17
m3 0.09 10.33
m4 0.03 11.67
m5 0.0 0
percent out of 10000 tries
**** BM 2 SD
f0 99.31 47.20
f1 85.44 20.85
m0 0.69 15.74
m1 0.4 14.82
m2 0.13 9.00
m3 0.01 19.00
m4 0.0 0
m5 0.0 0
percent out of 10000 tries
Unoriginal
2017-06-28, 08:58 AM
That's a pretty damning conclusion. I'll certainly keep it in mind if I'm in a campaign where there are only 2 players, we both have to play the same subclass of fighter with the same gear, we top out at level 5, and all our opponents are apes.
Hey now, don't be mean. It could be a Mario and Luigi vs Donkey Kong & family campaign
Lol. Now run it again with a half orc champ with a greataxe.
Just change fighter_rapier() to fit your need.
How is one sup dice making such a difference?
2d8+7 has 23% chance of doing 19 damage, enough to 1-shot the ape.
Unoriginal
2017-06-28, 09:07 AM
I have to ask, though: why apes?
Lombra
2017-06-28, 09:10 AM
Putting more work in what I spitted before I came to some conclusions with this method: conpare tha extra damage given by Superiority Dies versus the damage given by the Superior Critical feature, both characters are built the same for simplicity's sake, (I personally like ranged battlemasters):
The adventuring day will consist of 6 encounters with an average of 3 rounds of fighting for each encounter, this is important because it will determinate the amount of attacks done at a given level. There will be a short rest after every 2 encounters. Both fighters are half-orcs wielding a greataxe to maximize damage.
LEVEL 3
6fights ×3ounds ×1attack +3 attacks from action surge
= 21 attacks.
freshly gained archetypes, the battlemaster starts out very strong with 12d8 to unload diring the adventuring day, while the champion crits 1 more time than his partner, for a mere 2d12 extra damage across the whole adventuring day.
BM=12d8 ~54
Champ=2d12 ~13 BM >>>> Champ
LEVEL 5
39 attacks
Extra attack, more chances to crit. Both characters pick GWM at 4. The battlemaster has still 12d8 of extra damage, the champion will land two mpre hits than his partner for 2d12 each plus a bonus action attack each for a total of 6d12+2×STR (which is 3 if we assume point-buy)
BM=12d8 ~54
Champ=6d12+6 ~45 BM > Champ
LEVEL 10
Improved superiority (d10 Sup. Dice) and an extra Sup. Die
BM=15d10 ~83
Champ=6d12+10 (increasing strength with two ASIs) ~49
BM >> Champ
LEVEL 11
57 attacks.
Moar attacks, moar crits. The champion crits 3 additional times compared to his partner for a total of 6d12+3d12+15 (GWM extra attacks)
BM= 15d10 ~83
Champ= 9d12+15 ~74 BM > Champ
LEVEL 15
Still 57 attacks.
More superiority dies for the BM, better crit range for the champion, the champion scores an additional 6 critical hits compared to the BM, for a total of 12d12+6d12+30 damage and gets a .8 chance of another critical hit with those GWM bonus attacks. It exploded. Let's add the martial adept feat to give more superiority dies to the BM.
BM=21d10 ~116
Champ=18d12+30 (close to 20d12+30) ~147 BM < Champ
LEVEL 18
d12 superiority and 3 extra attacks from action surge (gained at level 17)
BM=21d12 ~137
Champ=20d12+30 ~160 BM < Champ
LEVEL 20
Attacks are raining from the sky: 78 total
The champ is critting an additional 8 times over the BM, giving 16d12+8d12+40+2d12(critical from the bonis attack)
BM= 21d12 ~136
Champ=26d12+40 ~209 BM << Champ
The champion starts lame and grows like a slow exponential curve, while the battlemaster keeps the damage with a linear increase.
The battlemaster obviously has nice riders on top of damage as a plus, while the champion really has not much besides damage itself. Level 3 is cathastropic, at level 10 it seems like things will never go well for the Champion but from level 15 it works. Yeah it just works too late to be considered cool.
Please address any mistakes or foggy spots. Heah there are typos but I'm on my phone so suffer with me
mgshamster
2017-06-28, 09:12 AM
I look forward for your in-depth analysis.
I've done some more analysis this morning (instead of doing work like I should). My previous analysis was off. I will be looking into it more carefully and taking careful notes over the next week, and then post it for peer review.
Maybe I'll even start a new thread just to see if I can make mephnick's head explode. :)
Mongobear
2017-06-28, 09:56 AM
replying to both the sim results as well as tossing some ideas for a 'fix' to make the Champion functional at earlier levels.
I always knew Champion took awhile to catch up to BM, but I never realized it took until 15th level, I figured at 11 theyd be about even and only get better from there.
I could be reading the code wrong, but are you using a SD on every attack the BM makes across 10000 fights? that's never possible and why the resulys say what they say, but I'm not a code guy so I'm probably wrong on this.
Also, using a SD for damage everytime is unrealistic, id wager with a GWM/SS build a BM would probably burn ~50% of those SD on Precision Attack just to make sure the -5/+10 actually benefits them.
As far as a fix for this, I think it is as simple as smushing some class features into earlier levels, and giving the class something to synergize with their critical hit chance boosting.
Level 3 - Imp Critical (19-20) and Remarkable Athlete (Changed to Bonus Proficiency in Acrobatics and Athletics or Expertise if already proficient, increased Jump distances, and +5 ft. of movement speed.)
Level 7 - Extra Fighting Style
Level 10 - Something akin to Brutal Critical, or maybe "Add Fighter level to damage of your critical hits"
Level 15 - Survivor
Level 18 - Superior Critical (18-20)
Malifice
2017-06-28, 10:28 AM
Did I miss something?
Try it again assuming five rounds per combat. Around about 30 rounds of combat per adventuring day should be the norm.
Also the fighter gets two action surges per rest at higher level.
That add up to about 36 attack actions. Multiply this by four adds to approximately 140 attack rolls per adventuring day at 20th. The champion scores approximately an extra 14 critical hits during this adventuring day (many more if he has a reliable source of advantage - advantage and in 18 -20 critical hit chance pretty much gives him a roughly 30% chance of every attack being a critical hit, resulting in him scoring about 30 critical hits).
Presuming he is reasonably optimised he is a half orc swinging a great axe. 28d12 per day. Which is reasonably on par with what the battlemaster is doing.
He obviously benefits from advantage more then the battle master. He also benefits when there are less short rests than expected during the day, or more encounters. If the party is unable to get short rests then he is clearly dominant.
The fact he is also regenerating at this level clearly helps also.
Whether the champion is mechanically good or not really comes down to the meta of the the campaign. If your DM allows you to routinely get short adventuring days, or short rest at will then you should probably (mechanically speaking) avoid it.
Of course every table has that one guy who just wants to rock, up roll dice and smash stuff. He isn't interested in options he just wants to kick things teeth in, and keep it simple. For that guy this is the perfect class. The fact it synergises so well with barbarian (Reckless attack being a reliable source of advantage, and brutal critical pairing nicely with the increased critical range) isn't just a coincidence.
mephnick
2017-06-28, 10:56 AM
You ever try.... not clicking on threads you aren't interested in?
Like half the god damn threads here are about the Champion and I'm trying to waste time at work.
mgshamster
2017-06-28, 11:00 AM
Try it again assuming five rounds per combat. Around about 30 rounds of combat per adventuring day should be the norm.
I've been assuming 24 per day, but I'm going to start using 30.
Also, I think we should stop saying 6-8 combats per day. Instead, we should say 25-30 rounds of com at per adventuring day, broken up into various combats (easy-deadly), with planned short rests in between.
That way, if people want to do 1-3 combats a day, they can plan for those combats to be very long (10-30 rounds).
"I only do one combat per day."
"How many rounds is it?"
"Oh, about four to six."
"That's way too short. If you're only doing so few combats, they need to be much longer; add a lot more combatants to bring the number of rounds up into the mid 20 range. Or better yet, split those bad guys up into multiple smaller combats throughout the day so your short rest guys can refresh."
Thoughts?
Daphne
2017-06-28, 11:36 AM
Try it again assuming five rounds per combat. Around about 30 rounds of combat per adventuring day should be the norm.
Agreed, most of the time combat last five or even six rounds on my table.
Presuming he is reasonably optimised he is a half orc swinging a great axe. 28d12 per day. Which is reasonably on par with what the battlemaster is doing.
Greatswords and Mauls are actually better because of GWF most of the time, even for Half-Orcs. Greataxe is only good against enemies with very high AC.
Lombra
2017-06-28, 11:50 AM
I did another thread to talk about the adventuring day damage and stuff so that we can keep this thread on the python system, hop in if you are intreasted! (It would be cool if someone could provide average damage of great weapon fighting over there because I can't do it)
Garresh
2017-06-28, 12:06 PM
I'm seriously tempted to do one in C# for kobolds and differing variables.
Garresh
2017-06-28, 12:11 PM
I did another thread to talk about the adventuring day damage and stuff so that we can keep this thread on the python system, hop in if you are intreasted! (It would be cool if someone could provide average damage of great weapon fighting over there because I can't do it)
Really? That's like the easiest thing to model. Just call your rand function again if it generates a 1 or a 2...
Hell you can even solve for average for each die. Here I'll show you for d6.
d6 normal spread: 1, 2, 3, 4, 5, 6
Add up to 21 and divide by 6 for 3.5
d6 gwf spread: 3.5, 3.5, 3, 4, 5, 6
add up to 25 and divide by 6 for 4.16~
Don't call it recursively. It only procs once.
You ever try.... not clicking on threads you aren't interested in?
The point stands. There is no reason to discourage people from playing the Champion. It is irrelevant how many statistical analyses are done that prove Champion does 1 or 2 less average damage per round than Battle Master or Eldritch Knight. It is an insignificant amount that does not impact game play. Players who want fiddly bits of spending class resources to do stuff are welcome to do so. Players who don't want to do that can play Champion instead, and if anyone has a problem with that get over it.
mgshamster
2017-06-28, 12:44 PM
The point stands. There is no reason to discourage people from playing the Champion. It is irrelevant how many statistical analyses are done that prove Champion does 1 or 2 less average damage per round than Battle Master or Eldritch Knight. It is an insignificant amount that does not impact game play. Players who want fiddly bits of spending class resources to do stuff are welcome to do so. Players who don't want to do that can play Champion instead, and if anyone has a problem with that get over it.
I agree.
Despite all my analysis showing the champ doesn't perform as well as the BM, and even analysis that shows TWF lags behind, I still love my Lizardfolk TWF champion.
Also, I have several character ideas that just feel wrong with anything except a champ.
alchahest
2017-06-28, 12:55 PM
Quick question; I've got code blindness - does the code account for BMs turning near misses into hits?
Lombra
2017-06-28, 01:01 PM
Really? That's like the easiest thing to model. Just call your rand function again if it generates a 1 or a 2...
Hell you can even solve for average for each die. Here I'll show you for d6.
d6 normal spread: 1, 2, 3, 4, 5, 6
Add up to 21 and divide by 6 for 3.5
d6 gwf spread: 3.5, 3.5, 3, 4, 5, 6
add up to 25 and divide by 6 for 4.16~
Don't call it recursively. It only procs once.
Well all I have is a piece of paper and no C# knowledge, so no models for me xD but as I understand it now that you described it I could calculate it by adding to 3.5 (the normal average) 3.5/3 which is the average times the probability to reroll? Boy I derped so hard. Thank you.
KorvinStarmast
2017-06-28, 01:58 PM
I'll certainly keep it in mind if I'm in a campaign where there are only 2 players, we both have to play the same subclass of fighter with the same gear, we top out at level 5, and all our opponents are apes. Let's not confuse D&D and WoW capture the flag runs ... :smallbiggrin:
There is no reason to discourage people from playing the Champion. It is irrelevant how many statistical analyses are done that prove Champion does 1 or 2 less average damage per round than Battle Master or Eldritch Knight. It is an insignificant amount that does not impact game play. Players who want fiddly bits of spending class resources to do stuff are welcome to do so. Players who don't want to do that can play Champion instead, and if anyone has a problem with that get over it. Yeah. :smallcool:
Elderand
2017-06-28, 03:12 PM
The point stands. There is no reason to discourage people from playing the Champion. It is irrelevant how many statistical analyses are done that prove Champion does 1 or 2 less average damage per round than Battle Master or Eldritch Knight. It is an insignificant amount that does not impact game play. Players who want fiddly bits of spending class resources to do stuff are welcome to do so. Players who don't want to do that can play Champion instead, and if anyone has a problem with that get over it.
According to my own calculation the damage difference between a champion and a battlemaster is this:
Level
Difference
1
0
2
0
3
2.49
4
2.58
5
2.12
6
2.18
7
2.95
8
2.99
9
-8.23
10
-6.97
11
3.02
12
3.02
13
3.02
14
3.02
15
0.43
16
0.43
17
3.15
18
3.83
19
3.83
20
2.83
Yes, strangely enough at level 9 and 10 we see the largest difference in dpr.....in favor of the champion.
Quick question; I've got code blindness - does the code account for BMs turning near misses into hits?
The SD is only used on crits and riposte.
The point stands. There is no reason to discourage people from playing the Champion. It is irrelevant how many statistical analyses are done that prove Champion does 1 or 2 less average damage per round than Battle Master or Eldritch Knight. It is an insignificant amount that does not impact game play.
There's no reason to keep spreading the "bestest that keeps on going" false belief. Players have to pick a weaker choice knowingly.
Every step of the way, zealots of the Cognitive Dissonnance Church have stated dipping 3 levels was the bestest. Every step of the way, they find new ways to be misunderstood:
- yeah but... at level 20...
- yeah but... as a half-orc...
- yeah but... I have advantage...
- yeah but... I can fight 100 rounds and catch up...
And now it's "irrelevant" and "insignificant".
Does that mean we are done with the champion lie?
You have the tools to explore which encounters work well for champions.
- you don't like apes? pick worgs or harpies.
- you'd rather have level 11? update the fighters.
- you want a bigger party? add more characters.
- you want reckless GWM? refactor the code.
If you don't believe the probalistic analysis that were done over the year, here's your chance to find an actual counter-example where somehow champions come on top.
My bet is that it'll need to be a one-legged ham-loving half-orc dancing under the full moon.
alchahest
2017-06-28, 06:18 PM
The SD is only used on crits and riposte.
Thanks. Interestingly, when GWF/SS is used it can be a huge damage swing, turning a 0 into weapon+statmod+10. which can be a much larger increase than weapon die + superiority die being doubled, especially at lower levels.
it adds another layer certainly, that champion can't use.
I don't really feel that superiority dice are "Fiddly" any more than rolling weapon dice are fiddly or counting your hit dice are fiddly.
Thanks. Interestingly, when GWF/SS is used it can be a huge damage swing, turning a 0 into weapon+statmod+10. which can be a much larger increase than weapon die + superiority die being doubled, especially at lower levels.
Yeah. That's why GWM makes it worse for champion as demonstrated earlier this year.
scalyfreak
2017-06-28, 09:10 PM
These kinds of threads used to amuse me. Now they confuse me.
The champion must be deeply offensive in some way to be so hated that someone would spend hours of their day to try and talk others out of playing that class. Never mind who these others are, what kinds of character they prefer to play and how they prefer to play them... they must be saved at all costs from the inferior horror that is the Champion. And mathematical proof is the only way we can do that!
But we must do it, because it genuinely matters so very much to us what strangers on the internet do when they play TTRPGs with their friends. :smalltongue:
coolAlias
2017-06-28, 10:07 PM
But we must do it, because it genuinely matters so very much to us what strangers on the internet do when they play TTRPGs with their friends. :smalltongue:
Clearly you are in dire need of being saved from badwrongfun. Come, stay a while and listen. ;)
I still find the mathematical proofs to be quite interesting, even though I doubt that the vast majority of players would ever notice the difference in DPR.
The sample size of attack and damage rolls at any given session is so small as to be statistically irrelevant - any difference in DPR that is noticed will usually be chalked up to good/bad luck, but in my experience, no one is comparing at the table.
Usually we're all playing different classes, or at least different archetypes or using different weapons, so of course our damage is all going to be different. The only thing I really notice as far as DPR is whether we are winning or losing and my own contribution (or lack thereof) to that state of affairs.
KorvinStarmast
2017-06-28, 10:20 PM
But we must do it, because it genuinely matters so very much to us what strangers on the internet do when they play TTRPGs with their friends. :smalltongue: It must be done because apparently, a Champion once ate their snacks in Kindergarden.
Clearly you are in dire need of being saved from badwrongfun. Come, stay a while and listen. ;) . Only if you offer me pixips as my snack.
scalyfreak
2017-06-28, 10:34 PM
Clearly you are in dire need of being saved from badwrongfun. Come, stay a while and listen. ;)
Actually, I think the that fact I'm enjoying every moment of my badwrongfun is proof that I'm beyond saving, and you should just give up on me. :smalltongue:
I still find the mathematical proofs to be quite interesting, even though I doubt that the vast majority of players would ever notice the difference in DPR.
Which makes it all the more amusing, and confusing, that something no one would even notice has become so important.
That's the badwrongfun talking again. Ignore it. It's not quite house trained.
Oerlaf
2017-06-28, 11:42 PM
As I am a mathematician (I am a PhD student in maths), it pains me to see how probability estimates are often used.
Of course, the OP has done a good job to find a point estimation of some random variable, frankly speaking.
But that is not enough to make ANY conclusions. If you truly want to get some reasonable results, you should pick a threshold p (take p=0.95) for example and build a confidence interval for that variable using that threshold.
Having a confidence interval, you will be able to say: The probability of X lies within an interval (Y,Z) with probability p.
This result can lead to more logical and precise conclusions.
scalyfreak
2017-06-28, 11:48 PM
As I am a mathematician (I am a PhD student in maths), it pains me to see how probability estimates are often used.
Of course, the OP has done a good job to find a point estimation of some random variable, frankly speaking.
But that is not enough to make ANY conclusions.
Ah, yes. There is this minor detail too.
If you truly want to get some reasonable results, you should pick a threshold p (take p=0.95) for example and build a confidence interval for that variable using that threshold.
Having a confidence interval, you will be able to say: The probability of X lies within an interval (Y,Z) with probability p.
Care to detail how to reach that point? You certainly have the expertise on how to massage 10k results into such an interval.
The naive way would be to remove the "best" 5% results from BM and the "worst" 5% results from champion and see if it reverses the conclusion, but I certainly don't have a clue how significant that would be.
Educate me, please.
https://stackoverflow.com/questions/15033511/compute-a-confidence-interval-from-sample-data
This seems to do it for a normal distribution.
https://docs.scipy.org/doc/scipy/reference/stats.html#module-scipy.stats
Which distribution would be represent "hp left at the end of encounter", and how to compare the champion distribution to the BM distribution?
alchahest
2017-06-29, 02:04 AM
Some people like to discuss the theory behind classes. to walk into a thread you're not interested in and accuse people of telling other people they're not having fun the right way is boggling. Though if that's how you get your fun then who am I to tell you any different?
I don't think anyone should avoid playing the champion, especially if they have fun with it (and particularly if they like to just roll D20s without making mechanically supported choices - the argument that you can improvise applies to literally every class). That isn't a wrong way to have fun, it isn't a bad way to have fun. Enjoy it!
I don't know that you'll get any additional enjoyment from pretending that the champion has things it doesn't have, though. If the main argument to people that it's as well designed as classes that have more options and more things to do, more meaningful agency to interact with the game rules.. is that in some certain circumstances it comes out slightly ahead in damage over 60+ attack rolls, then you might be better off just playing the champ and letting us have our fun discussing the math?
There's no reason to keep spreading the "bestest that keeps on going" false belief. Players have to pick a weaker choice knowingly.
Every step of the way, zealots of the Cognitive Dissonnance Church have stated dipping 3 levels was the bestest. Every step of the way, they find new ways to be misunderstood:
- yeah but... at level 20...
- yeah but... as a half-orc...
- yeah but... I have advantage...
- yeah but... I can fight 100 rounds and catch up...
And now it's "irrelevant" and "insignificant".
Does that mean we are done with the champion lie?zealots of the Cognitive Dissonnance Church? It's called peer review.
encounters with lots of small hp monster as disadvantages for a class who's ability is big damage.
You can not use a encounter of which a character has a disadvantage to conclude he's inferior
single level, single monster, single encounter is not statisical evidence: it's anectodal evidence. All your '10000' hits do, is ignore a single mathematical equation (equivalent of d6 being subpar to d8, by proof of rolling them 10000 times instead of mathematical comparing 3.5 vs 4.5)
You can not use anecdotal evidence a sub-par build of to conclude he's inferior
D&D is played in a party.
You can't just say: if we look at a irrealistic senario, they don't come out on top, so in a realistic senario, they are subpar.
Each and everyone of those is a valid concern. Trying to handwave them away as irrelvant makes you the missionairy of the Cognitive Dissonnance Church.
Because, imagine, there's a specific build of champion out there that beats the BM to, but you handwave'd it away as irrelevant ... then what? how much is a proof worth that reaches the wrong conclusion?
Care to detail how to reach that point? You certainly have the expertise on how to massage 10k results into such an interval.
The naive way would be to remove the "best" 5% results from BM and the "worst" 5% results from champion and see if it reverses the conclusion, but I certainly don't have a clue how significant that would be.
Educate me, please.linky (http://www.wikihow.com/Calculate-Confidence-Interval).
Take what you want to know. For example, the number of rounds the fight takes
calculate the mean a.k.a. the average (sum it all up, divide by 10K)
calculate the standard deviation (for each value, substract the mean, and square it. take the average of those value (add them all up, and divide by 10K), and take the square root)
multiply the standard deviation with 1.96 and divide by 100
You can now say with 95% confidence that 2 level 5 champions will last {the mean - that last value} rounds, and {the mean + that last value} rounds against 4 apes, while 2 level 5 battle masters fight will last between {symelar calculation} rounds & {symelar calculation} rounds.
You can do the same, for instance, with hp left, etc ...
And for different levels, monster, etc ...
-----
My own acid test in the good old days of D&D 3.5 was to pick the first 5 monsters starting with an 'A' (in 5e that would mean pitting them against Aboleth, Angel (Deva), Animated object (armor), Ankheg and Azar.
tsotate
2017-06-29, 03:39 AM
Oh god who cares.
The next person that makes a thread about the Champion gets pistol-whipped.
But would they be pistol-whipped more effectively by a Champion or a Battlemaster? :belkar:
Sirdar
2017-06-29, 03:53 AM
But would they be pistol-whipped more effectively by a Champion or a Battlemaster? :belkar:
:smallsmile: Isn't the Gunslinger class by Matt Mercer designed to excel at this?
Oerlaf
2017-06-29, 03:59 AM
Care to detail how to reach that point? You certainly have the expertise on how to massage 10k results into such an interval.
The naive way would be to remove the "best" 5% results from BM and the "worst" 5% results from champion and see if it reverses the conclusion, but I certainly don't have a clue how significant that would be.
Educate me, please.
Nope. There is a simpe way. Since we do not know the exact probability, there is a mathematical formula proven that if Q is a quantile of standard normal variable of order p (for p=0.95 Q is approximately 1.645), then we can say that the real probability lies within interval (p1,p2) where
double p1 = (p+Q*Q/(2*N)-Q*Math.Sqrt(p*(1-p)/N+0.25*Q*Q/(N*N)))/(1+Q*Q/(N));
double p2 = (p+Q*Q/(2*N)+Q*Math.Sqrt(p*(1-p)/N+0.25*Q*Q/(N*N)))/(1+Q*Q/(N));.
I've tweeked the code so I can run it myself (can't run standard python here), and spout out the amount of rounds. (95% confidence interval, followed by average)
4 apes
1.07 to 3.54 (avr 2.31) // raw
1.03 to 3.50 (avr 2.27) // champion
0.82 to 3.23 (avr 2.02) // SD1 BM
0.84 to 3.11 (avr 1.98) // SD2 BM
5 apes
1.44 to 4.71 (avr 3.08) // raw
1.39 to 4.64 (avr 3.02) // champion
1.24 to 4.19 (avr 2.72) // SD1 BM
1.22 to 3.89 (avr 2.56) // SD2 BM
6 apes
1.82 to 6.27 (avr 4.05) // raw
1.79 to 6.16 (avr 3.98) // champion
1.58 to 5.55 (avr 3.57) // SD1 BM
1.50 to 5.03 (avr 3.27) // SD2 BM
Even against 6 apes: 3.98 vs 3.57 (or even 3.27) ? that's less then a round difference in actual combat. I call that a neglectable difference.
mgshamster
2017-06-29, 08:39 AM
I've tweeked the code so I can run it myself (can't run standard python here), and spout out the amount of rounds. (95% confidence interval, followed by average)
To make sure I understand that correctly:
The numbers are the amount of rounds it takes for either a champ or a BM to defeat that many apes by themselves, yes?
What's raw mean?
Also, can you do all 4 superiority dice? And then maybe do it again at higher levels when there's more dice and more attacks for the fighters?
I'm also curious to continue increasing the number of apes until we see a difference (if any) between the two classes.
Edit: And then, from there, what happens if we increase the crit threat range? At what crit threat range will the two be nearly identical?
(Sorry, I'm not a code person, so I can't do this myself).
The numbers are the amount of rounds it takes for either a champ or a BM to defeat that many apes by themselves, yes?The number of rounds, indeed. It's set up to be 2 warriors, X apes. ("raw", btw, is a fighter without abilities)
Also, can you do all 4 superiority dice? And then maybe do it again at higher levels when there's more dice and more attacks for the fighters?higher levels is that's gonna much more then a bit of tweeking (it's not my code, and I'm a native C/C++/java developer)
(you'll note not a huge difference for 4 SD, as the code only uses them on ripostes and crits)
---------------- (added 4SD fighter)
4 apes
1.07 to 3.54 (avr 2.31) // raw
1.03 to 3.50 (avr 2.27) // champion
0.82 to 3.23 (avr 2.02) // SD1 BM
0.84 to 3.11 (avr 1.98) // SD2 BM
0.83 to 3.10 (avr 1.97) // SD 4 BM
5 apes
1.44 to 4.71 (avr 3.08) // raw
1.39 to 4.64 (avr 3.02) // champion
1.24 to 4.19 (avr 2.72) // SD1 BM
1.22 to 3.89 (avr 2.56) // SD2 BM
1.23 to 3.83 (avr 2.53) // SD4 BM
6 apes
1.82 to 6.27 (avr 4.05) // raw
1.79 to 6.16 (avr 3.98) // champion
1.58 to 5.55 (avr 3.57) // SD1 BM
1.50 to 5.03 (avr 3.27) // SD2 BM
1.54 to 4.77 (avr 3.16) // SD4 BM
Less then a round difference in actual combat. I call that a neglectable difference.
---------------- from my acid test, I ran the test against animate armors (CR1) (as these, like apes, are also 2 attack monsters, so that didn't require much code tweeking)
1 armor
0.03 to 2.87 (avr 1.45) // raw
0.04 to 2.76 (avr 1.40) // champion
0.13 to 2.59 (avr 1.36) // SD1 BM
0.11 to 2.58 (avr 1.35) // SD2 BM
0.11 to 3.58 (avr 1.35) // SD 4 BM
2 armors
0.87 to 5.54 (avr 3.21) // raw
0.79 to 5.40 (avr 3.10) // champion
0.52 to 5.19 (avr 2.86) // SD1 BM
0.56 to 4.91 (avr 2.74) // SD2 BM
0.60 to 4.81 (avr 2.71) // SD4 BM
3 armors
1.88 to 8.63 (avr 5.26) // raw
1.76 to 8.33 (avr 5.05) // champion
1.58 to 7.91 (avr 4.75) // SD1 BM
1.50 to 7.37 (avr 4.44) // SD2 BM
1.52 to 6.90 (avr 4.21) // SD4 BM
Again, even with 4SD, the BM's average amount of rounds doesn't differ a full round.
It's called peer review.
encounters with lots of small hp monster as disadvantages for a class who's ability is big damage.
You can not use a encounter of which a character has a disadvantage to conclude he's inferior
single level, single monster, single encounter is not statisical evidence: it's anectodal evidence. All your '10000' hits do, is ignore a single mathematical equation (equivalent of d6 being subpar to d8, by proof of rolling them 10000 times instead of mathematical comparing 3.5 vs 4.5)
You can not use anecdotal evidence a sub-par build of to conclude he's inferior
D&D is played in a party.
You can't just say: if we look at a irrealistic senario, they don't come out on top, so in a realistic senario, they are subpar.
Each and everyone of those is a valid concern.
Now that's constructive criticism. You widen the issue with factual reasons instead of eaily debunked edge cases.
Fighter hits for ~10 and crits for ~15 against a 19 hp ape. Hit+crit will most likely kill, while hit+hit won't. It is not clear which side is disadvantaged by that. That concern should be easy to remove by making a giant 95hp / 10 attacks ape against a giant 98 hp / 4 attacks fighter.
Lombra is currently trying the mathematical comparison. It has been done before with clear results. It was refused by zealots as being "not real enough".
I posit that in every scenario, BM will do more damage than champion. This is excessively clear below level 10. Late game, barbarian 17 / fighter 3 will do better as champion. I will argue that rather than going La-La-Land on the bestest, fans should look for realistic cases where champion is close enough.
I've tweeked the code so I can run it myself (can't run standard python here), and spout out the amount of rounds. (95% confidence interval, followed by average)
4 apes
1.07 to 3.54 (avr 2.31) // raw
1.03 to 3.50 (avr 2.27) // champion
0.82 to 3.23 (avr 2.02) // SD1 BM
0.84 to 3.11 (avr 1.98) // SD2 BM
5 apes
1.44 to 4.71 (avr 3.08) // raw
1.39 to 4.64 (avr 3.02) // champion
1.24 to 4.19 (avr 2.72) // SD1 BM
1.22 to 3.89 (avr 2.56) // SD2 BM
6 apes
1.82 to 6.27 (avr 4.05) // raw
1.79 to 6.16 (avr 3.98) // champion
1.58 to 5.55 (avr 3.57) // SD1 BM
1.50 to 5.03 (avr 3.27) // SD2 BM
Even against 6 apes: 3.98 vs 3.57 (or even 3.27) ? that's less then a round difference in actual combat. I call that a neglectable difference.
{mgshamster: raw is without archetype features}
I would argue that hp left is a better measure, since that's what you'll carry onto the next fight. I don't expect it matters much.
Am I reading this correctly and 1 sigma is.. {6.16-3.98 / 1.645 ~ 1.3} and champion is about .5 sigma behind?
Unoriginal
2017-06-29, 09:37 AM
Seriously, though, why apes?
Seriously, though, why apes?
First alphabetical monster in the choices given. I used the http://donjon.bin.sh site for a medium/hard encounter. Ultimately, it is a simple creature that has no special behavior to code.
mgshamster
2017-06-29, 09:47 AM
Looking at the data:
If a BM uses 2 SD in the combat (saving some for the next combat), then there is not a significant difference between the BM, Champ, and Raw. All are within 1 round of each other. (Here, we are defining a significant difference as more than 1 round).
If the BM Nova's and uses all SD in one combat, then while they are still within a round of the champ, they are also one round ahead of Raw - a significant difference. The champ, however, is not one round ahead or raw and there is still insignificant differences between them.
Like bid, I'm also curious to how much HP is remaining at the end.
I would argue that hp left is a better measure, since that's what you'll carry onto the next fight. I don't expect it matters much.I would disagree. If this is anything as a realistic senario, then you're the tank, and the rest of the party is supposedly bussy doing something else (dealing with the BBEG or something). Every round you spend tanking the apes/armors/whatever, is a round you're stuck not helping your party in a different way.
Besides, there's a serious correlation between amount of rounds fighting, and damage received.
Am I reading this correctly and 1 sigma is.. {6.16-3.98 / 1.645 ~ 1.3} and champion is about .5 sigma behind?Their standard deviation is not really relevant, as that's an indication of how large the confidence interval is. So, it only indicates how consistant the amount of rounds is (a.k.a. lower standard deviation, means a lower change it deviates significantly from the average)
If you want, in reverse, it would make maximum minus minium, divided by two, times 100, divided by 1.96. So, the 6ape champ's st.dev was equal to 111, while the 6ape 2SDBM's st.dev was equal to 90.
If a BM uses 2 SD in the combat (saving some for the next combat), then there is not a significant difference between the BM, Champ, and Raw. All are within 1 round of each other. (Here, we are defining a significant difference as more than 1 round).Not even sure if 1 round is a significant difference.
average 3.27
95% are between 1.50 & 5.03
75% are between 2.24 & 4.30
40% are between 2.79 & 3.75 (so, 60% fall outside that range)
when there's a round difference, it's still very likely we're just talking between a slightly unlucky BM vs a slightly lucky one.
2, maybe 3 rounds, looks like a minimum for a significant difference.
JNAProductions
2017-06-29, 10:26 AM
So why are you using every single superiority dice in the fight? You typically need to make those last at LEAST two fights, before you get your short rest in.
So why are you using every single superiority dice in the fight? You typically need to make those last at LEAST two fights, before you get your short rest in.the 4 SD (superiority dice) and 2 SD don't differ too much; typically we seem to be comparing the 2 SD BM with the champ.
1.79 to 6.16 (avr 3.98) // champ
1.50 to 5.03 (avr 3.27) // SD2 BM
1.54 to 4.77 (avr 3.16) // SD4 BM
JNAProductions
2017-06-29, 10:31 AM
Okay. Fair enough.
Seriously, though, why apes?
First alphabetical monster in the choices given. I used the http://donjon.bin.sh site for a medium/hard encounter.
Actually I wonder what happened to
Here's a bit of python code to simulate encounters. It runs 100k fight between 2 fighters and 4 apes
And why the switch was made to 6.
(if it was because there wasn't that much of a difference anyway ... then, that would defeat the objectivity of even this anecdotical test: You can't alter the test when it you don't like the result ... )
coolAlias
2017-06-29, 10:42 AM
Actually I wonder what happened to
Here's a bit of python code to simulate encounters. It runs 100k fight between 2 fighters and 4 apes
And why the switch was made to 6.
(if it was because there wasn't that much of a difference anyway ... then, that would defeat the objectivity of even this anecdotical test: You can't alter the test when it you don't like the result ... )
I don't know, that seems to be how a lot of scientific studies are done these days - it must be legitimate! :P
warty goblin
2017-06-29, 10:58 AM
As I am a mathematician (I am a PhD student in maths), it pains me to see how probability estimates are often used.
Of course, the OP has done a good job to find a point estimation of some random variable, frankly speaking.
But that is not enough to make ANY conclusions. If you truly want to get some reasonable results, you should pick a threshold p (take p=0.95) for example and build a confidence interval for that variable using that threshold.
Having a confidence interval, you will be able to say: The probability of X lies within an interval (Y,Z) with probability p.
This result can lead to more logical and precise conclusions.
Having a confidence interval formula lets you say that. A specific confidence interval calculated from a particular data set does not mean that, since the true value is either in the interval or not and we have no way to know which it is. The correct interpretation of a frequentist confidence interval is that as the number of repetitions of the procedure approaches infinity, the proportion of intervals that contain the true value approaches 95%. Or whatever your confidence level is.
This is ignoring the equivalence between frequentist results and Bayes results under certain particular improper priors, in which case you can of course say that the parameter lies in the interval with 95% probability, so long as you believe the prior. Unless you're doing that though, it is a fundamental misunderstanding of a confidence interval to interpret it as a probability.
In this case it also doesn't seem to matter hugely; since we can generate an arbitrary amount of data from the model, we can make the CI as narrow as we desire. Ergo we'll eventually conclude that the Battlemaster and Champion are different, simply by reducing the standard error to something tiny, which entirely misses the more important point of whether or not the difference is practically significant.
mgshamster
2017-06-29, 11:53 AM
Not even sure if 1 round is a significant difference.
average 3.27
95% are between 1.50 & 5.03
75% are between 2.24 & 4.30
40% are between 2.79 & 3.75 (so, 60% fall outside that range)
when there's a round difference, it's still very likely we're just talking between a slightly unlucky BM vs a slightly lucky one.
2, maybe 3 rounds, looks like a minimum for a significant difference.
Fair point.
Thanks for doing all this. You, too, bid!
Easy_Lee
2017-06-29, 12:28 PM
As I've said before, champion was built to be the default archetype of the default class. He gets a series of +1s to base abilities, but no additional options. That appeals to some people, and very much doesn't to others. The champion is the opposite of a wizard.
One thing I think is worth noting: BM maneuvers are great, but they're the only good feature BMs get. Meanwhile, all of the champion features are useful. Additionally, BMs have to decide when it's best to use their features. For champions, there's no extra decision-making involved, because their features are always on.
alchahest
2017-06-29, 01:06 PM
As I've said before, champion was built to be the default archetype of the default class. He gets a series of +1s to base abilities, but no additional options. That appeals to some people, and very much doesn't to others. The champion is the opposite of a wizard.
One thing I think is worth noting: BM maneuvers are great, but they're the only good feature BMs get. Meanwhile, all of the champion features are useful. Additionally, BMs have to decide when it's best to use their features. For champions, there's no extra decision-making involved, because their features are always on.
All the math has shown that the Champ can, under certain conditions and with a long adventuring day, come up alongside and occasionally exceed damage done. They can't however, provide the versatility that BMs can without reducing damage done, and can't choose when to use it's class feature. "always on" is a weird selling point for a thing that is entirely based on random chance. They lack the ability to choose when to engage the mechanics of the subclass.
The class keeps up in damage, which is fine, and expected, but lacks in agency. and that's fine, if your player does not want to do anything other than make attacks. If that is fun for you, great! and if they do want to engage the mechanics in other ways, they have less agency to do so than other classes, who are capable of doing all the same improvisational things, but also have mechanical grounding for different attacks and abilities.
KorvinStarmast
2017-06-29, 03:07 PM
Ultimately, it is a simple creature that has no special behavior to code. Kind of like our dear old friend the Champion. :smallbiggrin: I enjoy playing mine.
Easy_Lee
2017-06-29, 03:13 PM
All the math has shown that the Champ can, under certain conditions and with a long adventuring day, come up alongside and occasionally exceed damage done. They can't however, provide the versatility that BMs can without reducing damage done, and can't choose when to use it's class feature. "always on" is a weird selling point for a thing that is entirely based on random chance. They lack the ability to choose when to engage the mechanics of the subclass.
The class keeps up in damage, which is fine, and expected, but lacks in agency. and that's fine, if your player does not want to do anything other than make attacks. If that is fun for you, great! and if they do want to engage the mechanics in other ways, they have less agency to do so than other classes, who are capable of doing all the same improvisational things, but also have mechanical grounding for different attacks and abilities.
The whole lacks agency argument assumes that extra options give the player extra actions. But do they?
Every class and archetype in 5e can operate in combat and contribute in a meaningful way. Having more options in combat is less important than having good options that apply to most circumstances. The champion who takes Archery and TWF, for example, can almost always do something in any given combat round. More damage always helps. He can certainly act any time an equivalent BM could act. So the champion isn't losing combat rounds (agency).
Out of combat, the champion has no additional options, but the only thing the BM can do out of combat that the champion can't is use one extra tool and size up an opponent. In other words, the only out of combat thing the BM can do is try to prepare for combat with a little more information. And it takes a full minute, so good luck finding the time.
What the champion does have out of combat is Remarkable Athlete, which applies to everything from breaking things to drinking contests to impromptu dance-offs - any physical activity. Remarkable Athlete is superior to Student of War and Know Your Enemy for most purposes.
So it's not fair to say that BMs have more agency. They aren't taking more actions than the Champion.
If your argument is that effectiveness = agency, then champion vs BM is irrelevant since neither is optimal for anything on their own.
alchahest
2017-06-29, 03:24 PM
The whole lacks agency argument assumes that extra options give the player extra actions. But do they?
Every class and archetype in 5e can operate in combat and contribute in a meaningful way. Having more options in combat is less important than having good options that apply to most circumstances. The champion who takes Archery and TWF, for example, can almost always do something in any given combat round. More damage always helps. He can certainly act any time an equivalent BM could act. So the champion isn't losing combat rounds (agency).
Out of combat, the champion has no additional options, but the only thing the BM can do out of combat that the champion can't is use one extra tool and size up an opponent. In other words, the only out of combat thing the BM can do is try to prepare for combat with a little more information. And it takes a full minute, so good luck finding the time.
What the champion does have out of combat is Remarkable Athlete, which applies to everything from breaking things to drinking contests to impromptu dance-offs - any physical activity. Remarkable Athlete is superior to Student of War and Know Your Enemy for most purposes.
So it's not fair to say that BMs have more agency. They aren't taking more actions than the Champion.
If your argument is that effectiveness = agency, then champion vs BM is irrelevant since neither is optimal for anything on their own.
I'm not sure how to respond to this - BM literally has maneuvers that add additional options in combat, additional ways of interacting with the mechanics that you can use when you decide you want to.
If both the champion and the BM want to knock someone over, the champion has to give up an attack to attempt it. The BM just adds a rider to a successful attack. Same thing if you want to push someone over a railing.
Champions aren't able to Riposte, ever. Champions can't hand out temp HP. Champions cannot grant movement to allies.
they exist in the same action economy, absolutely. but the champion has one lever to pull. the BM has a bevy of options.
Easy_Lee
2017-06-29, 03:28 PM
I'm not sure how to respond to this - BM literally has maneuvers that add additional options in combat, additional ways of interacting with the mechanics that you can use when you decide you want to.
If both the champion and the BM want to knock someone over, the champion has to give up an attack to attempt it. The BM just adds a rider to a successful attack. Same thing if you want to push someone over a railing.
Champions aren't able to Riposte, ever. Champions can't hand out temp HP. Champions cannot grant movement to allies.
they exist in the same action economy, absolutely. but the champion has one lever to pull. the BM has a bevy of options.
By that logic, wizards and bards have many times more agency than fighters, period, and fighters should never be played. Should one choose to play a fighter, BM and champion are irrelevant since EK has the most options.
Do you see my point?
Random Sanity
2017-06-29, 03:29 PM
I don't know, that seems to be how a lot of scientific studies are done these days - it must be legitimate! :P
If the data is being manipulated towards a preferred result, you can't call the study scientific without lying through your teeth.
alchahest
2017-06-29, 04:16 PM
By that logic, wizards and bards have many times more agency than fighters, period, and fighters should never be played. Should one choose to play a fighter, BM and champion are irrelevant since EK has the most options.
Do you see my point?
Wizards and Bards do have significantly more narrative agency. they literally have magical spells. Also my argument has never been to not play classes with less agency. In fact just one or two posts back I said that if it's fun to have fewer options then that's great. I am not saying not to play the class. I'm saying it lacks compared to other classes with mechanical interaction. So I don't see your point, because we agree it appears that the champion is alright to play. That doesn't make discussion of it's weaknesses moot, especially in a thread about the mechanical, mathematical comparison between it and other classes.
So I will say this in bold because you seem to skip over it whenever I say it: I am not saying not to play champion. I don't think anyone in this thread is saying that at all. This, and the other threads that have come up recently, are discussing the merits and flaws of the subclass.
Easy_Lee
2017-06-29, 04:30 PM
Wizards and Bards do have significantly more narrative agency. they literally have magical spells. Also my argument has never been to not play classes with less agency. In fact just one or two posts back I said that if it's fun to have fewer options then that's great. I am not saying not to play the class. I'm saying it lacks compared to other classes with mechanical interaction. So I don't see your point, because we agree it appears that the champion is alright to play. That doesn't make discussion of it's weaknesses moot, especially in a thread about the mechanical, mathematical comparison between it and other classes.
So I will say this in bold because you seem to skip over it whenever I say it: I am not saying not to play champion. I don't think anyone in this thread is saying that at all. This, and the other threads that have come up recently, are discussing the merits and flaws of the subclass.
Forgive me for reading the title of the thread, "statistical proof against champion." I had assumed from that title that someone was arguing against playing the champion...again.
But again, I don't think options and agency are the same thing. Agency is all about doing the most with your options. Just because a player has more options doesn't make those options better, and certainly doesn't mean the player will use those options. .
alchahest
2017-06-29, 04:38 PM
One can be against the mechanical implementation of the Champion without being against playing it. It's important that people play what they want to play.
as for agency - I'm speaking about mechanical agency, since there's no discussion to be had about improvisational items between classes, since they all have identical "Stats" when it comes to that. you want a wizard to swing from a chandelier? cool, do it!
But because D&D has so steeped in mechanics and rules, if we are discussing statistical proofs, it can stand to reason that we're discussing the mechanical interaction with the narrative. Which the champion lacks, when compared to every other subclass. One would think that they would gain something (like better damage!) to account for this lack of mechanical agency, but instead, as it turns out they have to meet a very specific set of requirements (half orc GWF, over 66 rounds of combat or what have you) to come out ahead.
I'm not saying the champ is garbage, just that it comes up short when it's main feature (more damage!) only lets it compete in that aspect, as opposed to the other subclasses, which can compete there and also offer so much more.
Easy_Lee
2017-06-29, 04:40 PM
One can be against the mechanical implementation of the Champion without being against playing it. It's important that people play what they want to play.
as for agency - I'm speaking about mechanical agency, since there's no discussion to be had about improvisational items between classes, since they all have identical "Stats" when it comes to that. you want a wizard to swing from a chandelier? cool, do it!
But because D&D has so steeped in mechanics and rules, if we are discussing statistical proofs, it can stand to reason that we're discussing the mechanical interaction with the narrative. Which the champion lacks, when compared to every other subclass. One would think that they would gain something (like better damage!) to account for this lack of mechanical agency, but instead, as it turns out they have to meet a very specific set of requirements (half orc GWF, over 66 rounds of combat or what have you) to come out ahead.
I'm not saying the champ is garbage, just that it comes up short when it's main feature (more damage!) only lets it compete in that aspect, as opposed to the other subclasses, which can compete there and also offer so much more.
The champion's main feature is not more damage. That's his first feature. His others include an extra fighting style, bonuses to checks, and free healing.
That's what I'm trying to explain. Those things are not trivial and absolutely come up in play. All of them.
alchahest
2017-06-29, 04:49 PM
they're not nothing, absolutely. It's part of why I'd never say the champion isn't worth playing (like PHB Beastmaster ranger, for example). But the power is all passive - and at the end of the day you never get the opportunity to make choices about your class features being used. You can't say "well this is a tough boss fight, so I'll roll crits with my bow, and make sure that I'm the one who is targeted by the enemies, so that my healing kicks in". Everything that the Champion has from the subclass is incidental, not in a trivial way, but in a non-interactional way - and most of it, he's unable to opt into when it's needed. It's not that it isn't powerful - you can play a champion and have fun making attacks all day long, and be good at it. As good as a Battlemaster in a few situations! And You'll be tougher at very high levels, absolutely. and toughness is a very endearing, fun concept to play with.
But any time you're interacting with the rules to do things outside of your melee or ranged attacks, you're doing so with a disadvantage (not Disadvantage the game mechanic) as compared to the battlemaster, who has options that allow him to not only do the thing, such as tripping, but to do so with additional damage.
I won't ever say not to play the champion, but I've also had a lot of experience playing both the champion and the battlemaster, and only one of the two felt like I was able to make mechanically supported tactical choices using my class abilities.
Easy_Lee
2017-06-29, 05:33 PM
Some people don't want more tactical choices, they just want abilities that work. If you think about it, fighters already have a lot of class abilities (action surge, second wind, indomitable, etc.) to manage on top of managing inventory, and your character's skills, and remembering what party members can do, and everything else. Then there's the fact that you can technically declare any action, forcing your DM to figure out how to roll it.
That's a lot to think about. For new players, it's overwhelming. For seasoned players who like a fighter's overall lack of bookkeeping, it can be annoying.
Careful tactical play isn't everyone's cup of tea. Some people who play door-kicking fighters like to play like door-kicking fighters. A champion lets players do that without putting them at an extreme disadvantage.
In a typical campaign, BM will do more damage. No one is arguing against that. But in many situations, a champion's features are more useful. The champion wins initiative more often, has an extra fighting style to work with when the main one is no good, and gets bonuses to a lot of different checks.
People who focus solely on damage miss all of those things.
(if it was because there wasn't that much of a difference anyway ... then, that would defeat the objectivity of even this anecdotical test: You can't alter the test when it you don't like the result ... )
You're free to try 4 and see how much you like the result. I think you'll be underwhelmed by its potential for conspiracy.
Make your homework instead of throwing cheap shots like the other 2 lazy wankers.
In a typical campaign, BM will do more damage. No one is arguing against that. But in many situations, a champion's features are more useful. The champion wins initiative more often, has an extra fighting style to work with when the main one is no good, and gets bonuses to a lot of different checks.
If people were arguing that, there'd be no contest. But there are still a few that hold onto the belief that a 3-level dip in champion is the best thing since sliced bread, that improved crit will boost your dpr significantly.
If there's something useful shown here, it's that improved crit does insignificant damage that has no measurable impact on the encounter result.
It also shows how quickly hp is drained. That single encounter drains more than 30 hp most of the times. You cannot risk a 4th battle, no matter the archetype, and must take a short rest. This kills the claim that champions can go on and on all day long.
If it kills the DPR argument and moves the appraisal into fixed maneuvers vs derring-do, that's good enough for me.
mgshamster
2017-06-29, 06:30 PM
This kills the claim that champions can go on and on all day long.
I think that argument comes from their 18th level ability. It's for high level play only.
Easy_Lee
2017-06-29, 06:51 PM
If people were arguing that, there'd be no contest. But there are still a few that hold onto the belief that a 3-level dip in champion is the best thing since sliced bread, that improved crit will boost your dpr significantly.
If there's something useful shown here, it's that improved crit does insignificant damage that has no measurable impact on the encounter result.
It also shows how quickly hp is drained. That single encounter drains more than 30 hp most of the times. You cannot risk a 4th battle, no matter the archetype, and must take a short rest. This kills the claim that champions can go on and on all day long.
If it kills the DPR argument and moves the appraisal into fixed maneuvers vs derring-do, that's good enough for me.
Actually, it doesn't show anything. This thread was theory and simulation, not actual play. And that's the problem. At the end of the day, no one on this forum has a meaningful amount of play evidence to point to. People have their own campaigns, and games they've heard about, but that's about it. A few people here may be involved enough with AL to pull data, but even that would be limited only to AL play.
The lack of real testing is important. Rocket scientist Wernher von Braun once said, “One good test is worth a thousand expert opinions.” The only actual game evidence I've heard of a Champion and Battle Master competing in the same group went a little something like this: it was awesome, they destroyed everything and we never had to worry about combat.
What all of the theory-crafting does show is that at most levels, the Battle Master isn't too far ahead of or behind the Champion. This isn't 3.5e, where disparity was so great that we needed tiers. Everyone is effective enough to fill their role.
As a result, I've lost my patience for this kind of nitpicking. No matter how many times I point out the usefulness of remarkable athlete and an extra fighting style, or how great it can be to pull crit chains at choice moments, some of you aren't going to listen. That's the reality. And that's fine.
Unoriginal
2017-06-29, 06:57 PM
they literally have magical spells.
That's not an argument for narrative agency, though.
alchahest
2017-06-29, 07:09 PM
Actually, it doesn't show anything. This thread was theory and simulation, not actual play. And that's the problem. At the end of the day, no one on this forum has a meaningful amount of play evidence to point to. People have their own campaigns, and games they've heard about, but that's about it. A few people here may be involved enough with AL to pull data, but even that would be limited only to AL play.
The lack of real testing is important. Rocket scientist Wernher von Braun once said, “One good test is worth a thousand expert opinions.” The only actual game evidence I've heard of a Champion and Battle Master competing in the same group went a little something like this: it was awesome, they destroyed everything and we never had to worry about combat.
What all of the theory-crafting does show is that at most levels, the Battle Master isn't too far ahead of or behind the Champion. This isn't 3.5e, where disparity was so great that we needed tiers. Everyone is effective enough to fill their role.
As a result, I've lost my patience for this kind of nitpicking. No matter how many times I point out the usefulness of remarkable athlete and an extra fighting style, or how great it can be to pull crit chains at choice moments, some of you aren't going to listen. That's the reality. And that's fine.
I've played champion. A lot. It's part of why I have so much to say about it. The problem is I can say "over three sessions my improved critical range didn't come up once" and "Remarkable Athlete's +2 didn't make as big a difference as the raging barbarian's advantage when we had to break down a door", and be told that my game wasn't right, or that it's anecdotal, or that it favored other characters. And the level 18 ability doesn't really factor in to most games, as most games aren't played at/past that level. Some are, and it's a great and powerful ability when it can be used. But no amount of anecdotal "in my game this happened" will work, because it may not match up with someone else's experience.
So we go to math. Because math is irrefutable, it's just numbers. Whether those numbers mean anything to anyone is entirely up to the participants of the conversation. None of this means Champion isn't worth playing, nobody is saying that. What's being said is that mathematically, the most commonly used element of the champion doesn't stack up. And that's okay. If you want to play that, not only are you welcome to, you are encouraged to. I'm not going to tell someone at my table "Don't play champ it doesn't keep up to Battlemaster" I'll say "cool, you are gonna crit some dudes so hard it's going to be awesome".
I'm not saying you have fun wrong, I'm not saying that you don't play the game right, I'm just saying that here, in this discussion thread about the mathematics of the mechanical side of the most commonly used element of the champion vs the most commonly used element of the battlemaster, that the champ comes up short, as it only offers near-parity, and doesn't have the same range of options.
Vogonjeltz
2017-06-29, 07:18 PM
Of course, critical boosting features work best on large dice weapons, generally polearms or two-handed swords
Right...
Using 1d12, no shield
It should be 2d6, re-roll 1s and 2s because that's the most effective, offensively. If they're level 10 then the Champion also gets another point of armor, presumably (unless they evened themselves out by taking Archery and GWFS for both ranged and melee effectiveness instead of trying to maximize melee).
6 fights / adventuring day with one short rest after every 2 fights, level 5 characters, 3 rounds/fight, 39 (+3 from action surge) attacks total:
1) 6 is on the low end, it's a 6-8 spectrum for expected.
2) 3 rounds is totally off base. A Hard fight will easily absorb 10 rounds when 0 resources are expended (i.e. auto-attack) and result in incapacitation of one or more characters in a party of four. The Battlemaster only has 4 dice per short rest meaning they can only hurry along a single fight, really. Regardless of the difficulty of that fight, that leaves two more Medium-Hard encounters before they get to rest again, so we're really looking at about 25 rounds of combat per rest, maybe 70-80 total.
In 80 rounds the Champion gains an average additional 8 critical hits thanks to the expanded crit range for +16d6 damage (re-rolling 1s/2s gives us +66.667 damage average), about 12.667 damage more than the Battlemaster over the course of the day, on average.
Downside, it's spread out and not on demand. Upside, it's more.
LordVonDerp
2017-06-29, 09:06 PM
Lol. Now run it again with a half orc champ with a greataxe.
Won't matter.
2) 3 rounds is totally off base. A Hard fight will easily absorb 10 rounds when 0 resources are expended (i.e. auto-attack) and result in incapacitation of one or more characters in a party of four. The Battlemaster only has 4 dice per short rest meaning they can only hurry along a single fight, really. Regardless of the difficulty of that fight, that leaves two more Medium-Hard encounters before they get to rest again, so we're really looking at about 25 rounds of combat per rest, maybe 70-80 total.
In 80 rounds the Champion gains an average additional 8 critical hits thanks to the expanded crit range for +16d6 damage (re-rolling 1s/2s gives us +66.667 damage average), about 12.667 damage more than the Battlemaster over the course of the day, on average.
In those 80 rounds (with 2 SR as you stated), the BM would use riposte for 2d6+5+1d8 ~ 16 each, 5 success is enough to cover the pidly 66 damage from extra crits, assuming you manage to miss with the 7 others. And as shown in the first post of the thread, the BM will have spent his whole hp pool, twice.
There goes to show how blissfull the zealots of the Cognitive Dissonance Church are.
LordVonDerp
2017-06-29, 09:20 PM
2) 3 rounds is totally off base. A Hard fight will easily absorb 10 rounds when 0 resources are expended (i.e. auto-attack) and result in incapacitation of one or more characters in a party of four.
10 is way too high.
The Battlemaster only has 4 dice per short rest meaning they can only hurry along a single fight, really. Regardless of the difficulty of that fight, that leaves two more Medium-Hard encounters before they get to rest again, so we're really looking at about 25 rounds of combat per rest, maybe 70-80 total.
1 maybe 2 hard per rest, tops, and 10 rounds is a really long and tedious fight. Also, a complete a complete lack of resource expenditure is a bit weird.
In 80 rounds the Champion gains an average additional 8 critical hits thanks to the expanded crit range for +16d6 damage (re-rolling 1s/2s gives us +66.667 damage average), about 12.667 damage more than the Battlemaster over the course of the day, on average.
80 rounds a day is way too high, even by your already high estimates.
Downside, it's spread out and not on demand. Upside, it's more.
Except it's actually less.
Cybren
2017-06-30, 12:04 AM
There's only really one argument against the champion, and it's "I don't want to play a champion". This argument is often given in disingenuous forms, like mathematical "proofs" or statistical comparisons to other subclasses in white room scenarios. They aren't relevant. What's relevant is: do you want to play a champion?
I've played champion. A lot. It's part of why I have so much to say about it. The problem is I can say "over three sessions my improved critical range didn't come up once" and "Remarkable Athlete's +2 didn't make as big a difference as the raging barbarian's advantage when we had to break down a door", and be told that my game wasn't right, or that it's anecdotal, or that it favored other characters. And the level 18 ability doesn't really factor in to most games, as most games aren't played at/past that level. Some are, and it's a great and powerful ability when it can be used. But no amount of anecdotal "in my game this happened" will work, because it may not match up with someone else's experience.
So we go to math. Because math is irrefutable, it's just numbers. Whether those numbers mean anything to anyone is entirely up to the participants of the conversation. None of this means Champion isn't worth playing, nobody is saying that. What's being said is that mathematically, the most commonly used element of the champion doesn't stack up. And that's okay. If you want to play that, not only are you welcome to, you are encouraged to. I'm not going to tell someone at my table "Don't play champ it doesn't keep up to Battlemaster" I'll say "cool, you are gonna crit some dudes so hard it's going to be awesome".
I'm not saying you have fun wrong, I'm not saying that you don't play the game right, I'm just saying that here, in this discussion thread about the mathematics of the mechanical side of the most commonly used element of the champion vs the most commonly used element of the battlemaster, that the champ comes up short, as it only offers near-parity, and doesn't have the same range of options.
I would argue as a general principle you can't make Improved Critical stack up, because mechanics predicated on that much variance are very unpopular with the sorts of players who already care about power optimization. The only requirement be that it works well enough that the people who are attracted to that variance don't feel like they picked a crummy trap option when they do so. (also it seems weird to argue that the battlemaster offers "more options". It has more buttons, sure, but just because their hot bar is filled doesn't mean they aren't just doing the same thing as the champion, which is... making attack rolls)
You're free to try 4 and see how much you like the result. I think you'll be underwhelmed by its potential for conspiracy.
Make your homework instead of throwing cheap shots like the other 2 lazy wankers.Firstly, the burden of proof lies with who declares not who denies. Lazliy pointing out flaws in a proof, without doing anything else, is sufficient to discredit it.
Secondly, I'm not quite sure how you missed a third of my results ...
I've tweeked the code so I can run it myself (can't run standard python here), and spout out the amount of rounds. (95% confidence interval, followed by average)
4 apes
1.07 to 3.54 (avr 2.31) // raw
1.03 to 3.50 (avr 2.27) // champion
0.82 to 3.23 (avr 2.02) // SD1 BM
0.84 to 3.11 (avr 1.98) // SD2 BM
5 apes
1.44 to 4.71 (avr 3.08) // raw
1.39 to 4.64 (avr 3.02) // champion
1.24 to 4.19 (avr 2.72) // SD1 BM
1.22 to 3.89 (avr 2.56) // SD2 BM
6 apes
1.82 to 6.27 (avr 4.05) // raw
1.79 to 6.16 (avr 3.98) // champion
1.58 to 5.55 (avr 3.57) // SD1 BM
1.50 to 5.03 (avr 3.27) // SD2 BM
Even against 6 apes: 3.98 vs 3.57 (or even 3.27) ? that's less then a round difference in actual combat. I call that a neglectable difference.
If people were arguing that, there'd be no contest. But there are still a few that hold onto the belief that a 3-level dip in champion is the best thing since sliced bread, that improved crit will boost your dpr significantly.
If there's something useful shown here, it's that improved crit does insignificant damage that has no measurable impact on the encounter result.... err ... how exactly? the two basic mechanics for high crit builds (advantage to increase the chance on a 19; and powerful crit damage (ex. high weapon dice, half orc, barbarian) - not just a single d8) are neither considered in in the test.
There goes to show how blissfull the zealots of the Cognitive Dissonance Church are. Maybe this is anecdotal evidence, but I will have to agree that a lot of my fights last more then 3 rounds as well.
Consider that your results are artificial - as it presumes the entire enemy cast gangs up on the tanks who don't use their native healing, nor have a cleric boosting them or a controller debuffing the others.
djreynolds
2017-06-30, 01:06 AM
The champion did not just materialize out of thin air.
They did not crawl out of some crazy primordial pool of chemicals that just got struck by lightning
They were designed by the makers of 5E
The makers understood that some people want to play uber powerful classes(some right out of the gate).... and some people do not.
If there are complaints go to SageAdvice and ask the designers, "why did you make this?"
I find the other classes silly and overpowered.
Moon Druid, paladin, warlock, mystics, certain cantrips (EB, GFB, BB) IMO tend to ruin the game.
I find the majority of charisma based casters are overpowered and multiclassing between them is silly and not worth my time. Paladin/warlock sounds cheesy, but if you can pull it off go for it.
But 5E, unlike, 1st and 2nd addition is allowing you to play want you want to play. You have freedom that others did not have before in previous additions. There are no alignment restrictions and the ability restrictions are minimal. To play a paladin in 1st edition, you had to roll well or you didn't play one.
You don't have to play a champion or a paladin, its a choice.
Do you want to play a grunt? Fighter works
You want to play a jet pilot with all the bells and whistles? Paladin, warlock, and mystic are right there.
You want to play an armored wizard, then play one.
Some players believe in concept vs crunch and optimization, and in 5E you can have both or none.
5E has really allowed us to play what we want to play and to be powerful, or to play a simple fighter.
ShneekeyTheLost
2017-06-30, 02:04 AM
Except the test doesn't show that the Champion is inferior to BM in every situation. It only shows that it is *slightly* inferior against a pack of apes. You are erroneously conflating success in this one particular type of contest to overall superiority in general. You are also operating in a vacuum, without taking into consideration any of the other players which will inevitably be involved as well. Which makes sense, there's FAR too many variables to realistically calculate out the entire statistics tree.
The fact that the variance is less than 5 DPR in most circumstances, even when you stack the deck in favor of BM, is pretty telling. In most games, you aren't going to notice that. Nor is it relevant.
Correct, the DPR variance presented in this thread is completely irrelevant. Why? Simple: enemies do not become impaired when damaged, therefore the only thing that matters is how many turns it takes to kill something. Since less than 5 DPR is *NOT* enough to tip that scale over, particularly not with the rest of the party chiming in, the point is entirely moot.
It also depends on what you want your Fighter to do. Damage output is not a Fighter's primary calling in most cases, it's the second. The primary calling of the Fighter is meatshield and aggro control. If you want a DPR beast, go Barbarian. Or Warlock. Or Paladin, for burst melee DPR.
Which is why the BM Fighter is superior in most cases, as long as he doesn't run out of maneuvers. Because he can lock enemies down and keep them from being effective while the rest of the party works together to kill them.
Goading Attack and Tripping Attack are key. The first punishes the opponent for not targeting you, the second keeps him from targeting anyone since he got knocked prone, and also gives everyone advantage on him. Situationally, Distracting Strike can give a Rogue advantage for sneak attack dice, Pushing Attack can be absolutely brutal if your allies have ways of punishing movement or have put some negative effects on the battlefield (Spike Growth, Cloud of Daggers, Stinking Cloud, Sleet Storm...). These further the Fighter's primary job, nothing the Champion gets does.
tl;dr: Pointless numbers are pointless, but Champion is still weak because it is pretending to be a Barbarian and failing.
Malifice
2017-06-30, 02:32 AM
Seriously, though, why apes?
Because he cherry picked the exact kind of monster that skews the test in favor of the BM.
Also, he just doesnt like apes. They steal things and throw poo.
Because he cherry picked the exact kind of monster that skews the test in favor of the BM.
Also, he just doesnt like apes. They steal things and throw poo.some other monsters in average combat rounds (after 10K fights), champion vs 1SD (2SD) BM
4 apes (CR1/2): 2.27 vs 2.02 (1.98)
4 bears (CR1/2) : 2.09 vs 1.94 (1.89)
2 animate armors (CR1) : 3.10 vs 2.86 (2.74)
1 awakened tree (CR2): 1.43 vs 1.40 (1.39)
1 manticore (CR 3): 1.55 vs 1.72 (1.68)
(as the program runs: 2 warriors vs these creatures, with the BM riposting as much as possible, or using SD on crits if he has them left)
note 1: armor, tree & manticore are picked as alphabetically first no-special-ability monster of their CR. brown bear is the alphabetically second monster (after ape)
note 2: there are all straight forward no-special-ability monsters. depending on their special abilities hard hitters like the champion might have advantage or disadvantage over multi-hitters like the riposte BM. (Azars deal damage to melee attackers as non-action (negative for multi-hitters); bandit captains get +2 AC as reaction vs 1 attack (negative for hard-hitters) )
Edit 2: amount of bears was wrong in the first run
djreynolds
2017-06-30, 03:58 AM
Except the test doesn't show that the Champion is inferior to BM in every situation. It only shows that it is *slightly* inferior against a pack of apes. You are erroneously conflating success in this one particular type of contest to overall superiority in general. You are also operating in a vacuum, without taking into consideration any of the other players which will inevitably be involved as well. Which makes sense, there's FAR too many variables to realistically calculate out the entire statistics tree.
The fact that the variance is less than 5 DPR in most circumstances, even when you stack the deck in favor of BM, is pretty telling. In most games, you aren't going to notice that. Nor is it relevant.
Correct, the DPR variance presented in this thread is completely irrelevant. Why? Simple: enemies do not become impaired when damaged, therefore the only thing that matters is how many turns it takes to kill something. Since less than 5 DPR is *NOT* enough to tip that scale over, particularly not with the rest of the party chiming in, the point is entirely moot.
It also depends on what you want your Fighter to do. Damage output is not a Fighter's primary calling in most cases, it's the second. The primary calling of the Fighter is meatshield and aggro control. If you want a DPR beast, go Barbarian. Or Warlock. Or Paladin, for burst melee DPR.
Which is why the BM Fighter is superior in most cases, as long as he doesn't run out of maneuvers. Because he can lock enemies down and keep them from being effective while the rest of the party works together to kill them.
Goading Attack and Tripping Attack are key. The first punishes the opponent for not targeting you, the second keeps him from targeting anyone since he got knocked prone, and also gives everyone advantage on him. Situationally, Distracting Strike can give a Rogue advantage for sneak attack dice, Pushing Attack can be absolutely brutal if your allies have ways of punishing movement or have put some negative effects on the battlefield (Spike Growth, Cloud of Daggers, Stinking Cloud, Sleet Storm...). These further the Fighter's primary job, nothing the Champion gets does.
tl;dr: Pointless numbers are pointless, but Champion is still weak because it is pretending to be a Barbarian and failing.
This is great post
But I like playing champions, I think players get "down" on themselves when the crits are do not pile up.
The champion IMO, wants to be 3.5E weapon master, but a slight boon is the crit range affects all weapons
Squat Nimbleness, brawny, expertise in athletics and either shield master or giving up an attack to shove, will increase the number of crits you land
I see the melee fighter, especially the champion as an American Football linebacker or even a fullback, you have multiple jobs on the battlefield but you are not dedicated to one aspect. As a linebacker sometimes you rush the quarterback, or stop the run, or drop back into coverage. As a fullback you make a hole or block. A very solid team player
But if you want DPR and glory.... go paladin, barbarian, or warlock.
But the negativity surrounding the champion, IMO, is due to players thinking Weapon Master and realizing critical hits don't stack up
But if you want DPR and glory.... go paladin, barbarian, or warlock.For all 4E's limitations, if you want DPR, go striker, not defender ... :smallwink:
djreynolds
2017-06-30, 04:19 AM
For all 4E's limitations, if you want DPR, go striker, not defender ... :smallwink:
Its painful because with some tweaking the champion could be a champion but alas its not.
I have run a few champions up to the higher levels. And you really have to work to be beneficial to the party.
There is a special charm of the champion archetype. Its the one archetype that every player wants to take a chance with and prove the he/she can survive, just me and my sword.
So if you think you a great gamer, take the champion challenge and see if you can survive
Zalabim
2017-06-30, 06:10 AM
I suspect I don't have to say anything more, but where's the fun in that?
as for agency - I'm speaking about mechanical agency, since there's no discussion to be had about improvisational items between classes, since they all have identical "Stats" when it comes to that. you want a wizard to swing from a chandelier? cool, do it!
They most certainly do not have the same stats for improvisational things. A melee fighter is probably going to improvise strength things better than a non-melee non-fighter, and if that improvisation is replacing attacks, then a fighter with extra attack is at least twice as good at it as a wizard without extra attack, and if that improvisation is just an Action, Fighters also get Action Surge. Fighters improvise quantifiably better than most.
But because D&D has so steeped in mechanics and rules, if we are discussing statistical proofs, it can stand to reason that we're discussing the mechanical interaction with the narrative. Which the champion lacks, when compared to every other subclass. One would think that they would gain something (like better damage!) to account for this lack of mechanical agency, but instead, as it turns out they have to meet a very specific set of requirements (half orc GWF, over 66 rounds of combat or what have you) to come out ahead.
When we're talknig statistical proofs, we're usually talking about statistically provable things, like damage and effective hit points, or time to kill and time to live, or other such things.
If people were arguing that, there'd be no contest. But there are still a few that hold onto the belief that a 3-level dip in champion is the best thing since sliced bread, that improved crit will boost your dpr significantly.
If there's something useful shown here, it's that improved crit does insignificant damage that has no measurable impact on the encounter result.
You didn't test the 3-level dip, so you can't kill the 3-level dip with this test. Sorry. I do think battle master is a better 3-level dip, FWIW.
1) 6 is on the low end, it's a 6-8 spectrum for expected.
For level 5, against Apes, 4 is a medium encounter and the game would expect 7 of those encounters a day. 5 is also a medium encounter but the game would only expect 5.6 of those encounters. 6 Apes is a Hard encounter and the game expects only 4.66~ of those encounters a day. The game does not expect a party to face 8 Hard encounters in a day. Maybe they could, but don't bet on it.
2) 3 rounds is totally off base. A Hard fight will easily absorb 10 rounds when 0 resources are expended (i.e. auto-attack) and result in incapacitation of one or more characters in a party of four. The Battlemaster only has 4 dice per short rest meaning they can only hurry along a single fight, really. Regardless of the difficulty of that fight, that leaves two more Medium-Hard encounters before they get to rest again, so we're really looking at about 25 rounds of combat per rest, maybe 70-80 total.
The 6 Ape fight in this thread is a Hard encounter and the two fighters using no resources end that fight in about 2 to 6 rounds. Their average is 4 rounds, and that's on the long end because they're facing multiple creatures and using single target attacks. If they were facing fewer, more dangerous creatures, the fight would end faster either way. Rounding up the Hard encounters per day to 5, they fight for an average of 20 rounds per day. 80 is right out.
some other monsters in average combat rounds (after 10K fights), champion vs 1SD (2SD) BM
4 apes (CR1/2): 2.27 vs 2.02 (1.98)
4 bears (CR1/2) : 2.09 vs 1.94 (1.89)
2 animate armors (CR1) : 3.10 vs 2.86 (2.74)
1 awakened tree (CR2): 1.43 vs 1.40 (1.39)
1 manticore (CR 3): 1.55 vs 1.72 (1.68)
(as the program runs: 2 warriors vs these creatures, with the BM riposting as much as possible, or using SD on crits if he has them left)
Of interest to me here, these encounters are:
(Apes or Bears): 400XP*2.5(for enemy group size compared to party size) = 1000 XP out of 1000-1500 XP for Medium
Animate Armors: 400XP*2 = 800 XP out of 500-1000 XP for Easy
Awakened Tree: 450XP*1.5 = 675 XP out of 500-1000 XP for Easy
Manticore: 700*1.5=1050 XP out of 1000-1500 XP for Medium.
The Manticore is the highest XP encounter, but has a shorter duration. The Apes are (entirely accidentally) the encounter that most favors the BM compared to the Champion. The two easy encounters are both the longest and the shortest encounters of the bunch.
Of interest to me here, these encounters are:
(Apes or Bears): 400XP*2.5(for enemy group size compared to party size) = 1000 XP out of 1000-1500 XP for Medium
Animate Armors: 400XP*2 = 800 XP out of 500-1000 XP for Easy
Awakened Tree: 450XP*1.5 = 675 XP out of 500-1000 XP for Easy
Manticore: 700*1.5=1050 XP out of 1000-1500 XP for Medium.
The Manticore is the highest XP encounter, but has a shorter duration. The Apes are (entirely accidentally) the encounter that most favors the BM compared to the Champion. The two easy encounters are both the longest and the shortest encounters of the bunch.If I were to venture a guess, I'd think this has more to do with requiring it requiring more enemies to kill.
(added 2 more:)
4 apes (CR1/2): 2.27 vs 2.02 (1.98)
4 bears (CR1/2) : 2.09 vs 1.94 (1.89)
2 animate armors (CR1) : 3.10 vs 2.86 (2.74)
1 awakened tree (CR2): 1.43 vs 1.40 (1.39)
1 manticore (CR 3): 1.55 vs 1.72 (1.68)
1 ettin (CR 4): 2.12 vs 1.93 (1.91)
1 barbed devil (CR 5): 3.70 vs 3.37 (312)
that last one, btw, has a confidence interval (95%) of
champ: 1.57 to 5.83 rounds
1SD BM: 1.31 to 5.43 rounds
2SD BM: 1.21 to 5.00 rounds
which would indicate we'd need to have a two or three round difference to have a noticable difference ... a 0.33 (or 0.58) difference in rounds is pretty neglectable.
Unoriginal
2017-06-30, 10:16 AM
Maybe a random selection of encounters of a given CR would help to gather more relevants data?
Firstly, the burden of proof lies with who declares not who denies. Lazliy pointing out flaws in a proof, without doing anything else, is sufficient to discredit it.
Considering your comical attempt at blaming it on conspiracy, I think you discredit yourself.
You were doing so well at finding counter-examples and reason why the difference doesn't matter, character assassination just cheapens your argument.
Coranhann
2017-06-30, 10:46 AM
Maybe a random selection of encounters of a given CR would help to gather more relevants data?
I think you should take 50/75 games with a group of 4 or 5 players, including a champion fighter, and record rolls, number of encounter per day, saved attack, players satisfaction, then do the same with a BM champion, and see how the stats look ?
... oh, sorry, I seem to have posted in the wrong thread.
Unoriginal
2017-06-30, 10:55 AM
I think you should take 50/75 games with a group of 4 or 5 players, including a champion fighter, and record rolls, number of encounter per day, saved attack, players satisfaction, then do the same with a BM champion, and see how the stats look ?
Systematic field research? You deviant
some other monsters in average combat rounds (after 10K fights), champion vs 1SD (2SD) BM
1 manticore (CR 3): 1.55 vs 1.72 (1.68)
Manticore barely has time for 1-2 hits before going down. Action surge on the first round is enough to kill it usually. Not a fight a DM would try, at least not without giving the manticore some range to soften up the fighters.
Since I suspect the result is because SD are never used that fast, let's try it:
**** raw fighter
f0 100.0 49.00
f1 99.98 39.63
m0 0.0 0
sd [10000]
rounds [ 0 2588 5515 1689 196 11 1]
percent out of 10000 tries
**** champion
f0 100.0 49.00
f1 99.95 39.87
m0 0.0 0
sd [10000]
rounds [ 0 2919 5419 1479 170 11 2]
percent out of 10000 tries
**** BM 1 SD
f0 100.0 49.00
f1 99.97 41.04
m0 0.0 0
sd [1556 7068 1376]
rounds [ 0 3931 5086 879 101 2 1]
percent out of 10000 tries
**** BM 2 SD
f0 100.0 49.00
f1 99.99 41.22
m0 0.0 0
sd [ 9 384 3393 4800 1414]
rounds [ 0 3985 5077 887 48 3]
percent out of 10000 tries
==== sd and rounds are in bins counting the occurence of 0, 1, 2, 3...
Indeed, no SD are used 13-14% of the time.
If you add a second manticore to make it a real fight instead of a speed bump, you return to the usual statistically insignficant gain for BM.
**** raw fighter
f0 98.5 46.23
f1 79.41 19.04
m0 1.5 23.03
m1 0.06 10.00
sd [10000]
rounds [ 0 0 8 1035 3667 3040 1174 540 307 137 67 17 5 2 1]
percent out of 10000 tries
**** champion
f0 98.93 46.52
f1 81.63 19.87
m0 1.07 25.36
m1 0.05 11.20
sd [10000]
rounds [ 0 0 36 1436 3978 2673 1010 447 220 125 53 17 3 2]
percent out of 10000 tries
**** BM 1 SD
f0 99.57 47.48
f1 87.02 21.37
m0 0.43 20.37
m1 0.0 0
sd [4577 5408 15]
rounds [ 0 0 201 2443 4028 2151 725 250 118 60 11 10 3]
percent out of 10000 tries
**** BM 2 SD
f0 99.86 48.05
f1 90.23 22.68
m0 0.14 22.29
m1 0.0 0
sd [ 930 3015 5747 293 15]
rounds [ 0 0 473 3395 3820 1620 475 150 47 15 4 0 1]
percent out of 10000 tries
Considering your comical attempt at blaming it on conspiracy, I think you discredit yourself.
You were doing so well at finding counter-examples and reason why the difference doesn't matter, character assassination just cheapens your argument.Firstly, "the burden of proof lies with who declares not who denies" has to do with proving a negative. It's not commical - it's both the basis of our justice system, scientific work (peer review), etc ...
Secondly. character assassination is a ad hominem. That's not what I'm doing. Oppositely, if there exists one or more example that disprove the theory, proves the theory is wrong. That's what I'm doing. (If I can provide one black swan, that proves that "all swans are white" is wrong). Especially, since all you brought is anecdotal evidence to the table, anecdotal evidence contradicting the claim is extremely damaging to the argument.
Thirdly "so well at finding counter-examples" ? DFQ you talking about? Those are simply the first creatures, alphabetically, of their challenge rating, that could be fit in your program. Seeing as that is almost exactly the same as what you have claimed you use to pick apes, I would be very careful to suggest that this method is subjective.
Indeed, no SD are used 13-14% of the time.
If you add a second manticore to make it a real fight instead of a speed bump, you return to the usual statistically insignficant gain for BM.I'm sorry ...
What, the actual f*** !?
You don't add the second manticore "to make it a real fight". A CR3 monster is conform with 4 CR 1/2s .
You add it, so the BM use their SDs more.
THAT is a clear example of tweeking the tests
(heck, the fact that the awakened tree encounter has the same problem - even more then the manticore, yet you don't care about that one, pretty much confirms this)
You didn't test the 3-level dip, so you can't kill the 3-level dip with this test. Sorry. I do think battle master is a better 3-level dip, FWIW.
Without barbarian 9 and brutal critical, there's no improvement to critical damage. I suspect most 6/3 MC would behave the same way as a pure fighter 9. I don't think the increased hp pool would change results, but it's worth the try.
As for the barb 17 / 3, I believe the average gain from 3x crit on 19 beats SD quickly. I don't remember the necessary conditions I used to calculate the average, though.
You don't add the second manticore "to make it a real fight". A CR3 monster is conform with 4 CR 1/2s .
You mean you, as a DM, would make a speedbump fight?
How about the fighters don't waste their action surge on something that easy?
How about the manticore had a single round of ranged tail spikes before being engaged in melee?
In both cases, it comes down to this:
**** raw fighter
f0 100.0 48.99
f1 99.73 32.94
m0 0.0 0
sd [10000]
rounds [ 0 0 2478 5590 1678 233 18 3]
percent out of 10000 tries
**** champion
f0 100.0 48.98
f1 99.75 33.60
m0 0.0 0
sd [10000]
rounds [ 0 0 2851 5476 1476 176 20 0 0 1]
percent out of 10000 tries
**** BM 1 SD
f0 100.0 48.99
f1 99.79 35.17
m0 0.0 0
sd [1902 7687 411]
rounds [ 0 104 4588 4336 880 81 11]
percent out of 10000 tries
**** BM 2 SD
f0 100.0 49.00
f1 99.96 35.80
m0 0.0 0
sd [ 29 838 5647 3113 373]
rounds [ 0 158 5225 3955 617 43 2]
percent out of 10000 tries
Now the fight has a good chance of reaching 3 rounds and making it interesting.
Surely as a DM you would make sure the fight is a challenge.
JNAProductions
2017-06-30, 12:05 PM
If the math shows you're incorrect, you don't change the formula, you change your opinion.
Or at least, you should.
AmmunRa
2017-06-30, 12:17 PM
Still reading through the code, just got to say I think this is the coolest thread I've seen here in a long time!
How about the fighters don't waste their action surge on something that easy?Oh? Good idea. Lets metagame in the fighters knowing how challenging monster are.
Seriously, you honestly thought that's a good proposal ???
The manticore was pitted against the 2 fighters, because people were saying you cherry picked the apes, so I pitted the fighters against other monsters (4 CR 1/2, 2 CR1, 1 CR2+). Objective encounters.
Your blatant attempt to modify the one encounter that champions got ahead, even ignoring that each and every result so far doesn't even differ to any significant degree ... even retroactively trying to justify it by saying the fighters can metagame ... ?
Well, that's just ... beyond words.
You mean you, as a DM, would make a speedbump fight?
Surely as a DM you would make sure the fight is a challenge. As DM, tweeking your encounters to your party is (arguably) good form (esp, if you add moments so certain PCS can get spotlight).
As analysist, tweeking your encounter is at best ignorant, at worst malicious.
If the math shows you're incorrect, you don't change the formula, you change your opinion.
Or at least, you should. +1
Unoriginal
2017-06-30, 12:31 PM
You mean you, as a DM, would make a speedbump fight?
[...]
Now the fight has a good chance of reaching 3 rounds and making it interesting.
Surely as a DM you would make sure the fight is a challenge.
Hold on, hold on.
I'm sorry, what?
Since when making cool fights is the purpose of this exercise?
I thought it was about determining facts, not making stuff look awesome. What kind of chemist would go "I've added copper to our new fuel. It serves no purpose, but the green fire will be sick as hell when we test it"?
Also, I don't know for you, but as a DM I wouldn't make the PCs fight groups of apes of various sizes three times back-to-(silver)back or more. Unless if I was being facetious and my players didn't mind me monkeying around.
THAT is a clear example of tweeking the tests)
That's a clear proof you don't understand the limitations of the simulation. You were testing the artifacts of the simulation.
The test must be a representative sample of a short-rest's worth of encounters. The manticore failed that for multiple reasons. The awakened tree prolly had the same issues, but why repeat the obvious.
- manticore starts in melee range, dropping its CR since it cannot use its tail spikes
- fighters using their action surge in an encounter they'd have to repeat 4 times before needing some rest
- lack of handling of SD for tripping or other early uses.
The first step in statistical analysis is validating the test case against the simulation's limitations. Failing to respect that yields lies and damn statistics.
Now here's what happens when I modify the code to make consecutive fights, each with a single monster.
**** raw fighter
f0 99.79 47.50
f1 86.42 18.97
m0 0.21 18.95
m1 0.0 0
sd [10000]
rounds [ 0 0 0 9 1065 3856 3204 1201 403 164 57 30 7 3 1]
percent out of 10000 tries
**** champion
f0 99.81 47.68
f1 88.36 19.36
m0 0.19 13.11
m1 0.0 0
sd [10000]
rounds [ 0 0 0 48 1375 4064 2904 1058 334 130 56 22 7 2]
percent out of 10000 tries
**** BM 1 SD
f0 99.87 48.17
f1 90.7 20.93
m0 0.13 15.08
m1 0.0 0
sd [4462 5529 9]
rounds [ 0 0 0 211 2504 4044 2282 658 199 73 20 6 2 1]
percent out of 10000 tries
**** BM 2 SD
f0 99.98 48.51
f1 93.57 22.59
m0 0.02 6.00
m1 0.0 0
sd [ 773 3142 5879 201 5]
rounds [ 0 0 9 594 3374 3764 1648 480 97 28 5 0 1]
percent out of 10000 tries
The fighters always kill m1 on the first fight, and then have to handle the second encounter without action surge.
That's a clear proof you don't understand the limitations of the simulation. You were testing the artifacts of the simulation.
- manticore starts in melee range, dropping its CR since it cannot use its tail spikes
- fighters using their action surge in an encounter they'd have to repeat 4 times before needing some rest
- lack of handling of SD for tripping or other early uses.No. Not artifacts. Specified behavior
The map
If it's OK that apes have no trouble surrounding the attacked fighter (even if fighter 2 gets between the first ape and the other three), then it's OK that the manticore starts within mellee or single movement range.
A player's behavior
The players behavior was that use their action surge instantly, and use their SDs on riposte or use it on crits.
As they got no reason to change their Modus Operandi, you can't argue that "if they metagame, they are better"
As for the side effect of the SD?
They can only do it if they haven't used it on riposte yet
AND they only use it on crit (re. player behavior)
AND it has to be useful (not disarming attack, manouvring attack, pushing attack, trip right before the monster's turn; or goading attack by the wounded fighter)
AND the monster usually gets a save if it's a penalty
AND it has to has to alter the amount of rounds before it shows up in the statistics,
... I really wouldn't hold my breath.
===============
Sorry mate, but still. No dice.
And seriously ... what your obsession is to throw away all your credibility, just because there's one senario when the champ gets slightly ahead, is beyond me.
Look at those numbers. Your own numbers
rounds [ 0 0 0 48 1375 4064 2904 1058 334 130 56 22 7 2]
rounds [ 0 0 0 211 2504 4044 2282 658 199 73 20 6 2 1]
You happy? Squirm Squirm Squirm. Modify the test where the champion came out on top. Lets add another monster. Lets change action point Significant difference between the two? still nothing.
Takes 4-6 turns to beat them (supposing your array was 0-based), depending on your luck.
Look at those numbers. Your own numbers
rounds [ 0 0 0 48 1375 4064 2904 1058 334 130 56 22 7 2]
rounds [ 0 0 0 211 2504 4044 2282 658 199 73 20 6 2 1]
Yes, the difference is negligible and not statistically significant. Who cares.
Oh wait... are you actually happy because 4064 is bigger than 4044? That would be a huge brainfart.
Every sample tried behaves the same way. Everything stays coherent. If the difference was significant, it might be worth improving the code to properly handle multiple encounters, explore level 11 and more, add ranged delay, anything to widen the validity of the results. As it stands, the archetype has an insignificant impact on the fighter battles.
You come up with a sample (manticore) where the behavior was different from the rest. That might have been a sign that there was a bug in the code that made the whole simulation invalid. It was worth exploring.
I explained why it was not a representative sample of a short-rest's worth of encounters. I even demonstrated the artifact vanishes when we remove the limitation of the simulation by making 2 consecutive encounters.
Are you going to argue a single fight where a single party member takes 15 damage is enough to stop and take a short rest?
Do you call that credible?
Yes, the difference is negligible and not statistically significant. Who cares.
Oh wait... are you actually happy because 4064 is bigger than 4044? That would be a huge brainfart.Is that what you saw? that 4064 is bigger then 4044? I didn't. I saw that 40% of the combats last 5th, and another 40% lies in the adjacent rounds.
Nobody who knows anything about these types of calculations gives two ducks about that 20 point difference.
Oh wait... are you actually happy because 4064 is bigger than 4044?
In the end, though, I don't care what the values are. They are to be discarded as they are corrupted by your need to bash the champion.
To quote YOU
> Seriously, though, why apes?
First alphabetical monster in the choices given. I used the http://donjon.bin.sh site for a medium/hard encounter. Ultimately, it is a simple creature that has no special behavior to code.
To quote http://donjon.bin.sh/
5e Encounter Size Calculator
for a medium encounter for 2 characters of 5th level
CR 3: 1
for a deadly encounter for 2 characters of 5th level
CR 3: 2
1 Manticore, is a medium encounter of the alphabetically first CR 3 creature that doesn't require special behavior to code.
... but then, the champion did, with an insignificant about, slightly better then the battlemaster.
HOLD THE PRESS. Screw medium encounters. who said medium? We need deadly ones to see how the champion fairs vs the battlemaster. Who said medium? Clearly some zealot of the churge of Cognitive dissonance, I'm sure !
We need to change this ! Lets see how we can let the battlemaster use his SD more! Maybe we should allow the fighters to metagame !
You think you fixed an anomaly? You didn't: you changed your test so the results are conform your expectations.
You think it's better now? more realistic, doesn't mean better. To take one simple example: you didn't chose to change how a champion could trigger opportunity attacks against 4 apes (who might not start in melee reach either; and those and seeing as BMs use their reaction on riposte, they might not get those OAs) - you chose to change the rules of the fights, because you deemed the champion being ever-so-slighty better, to be something that needed to be fixed.
DM > champ.
If not, anomaly
If anomaly, improve test so anomaly no longer presides
Hey look, DM > champ
When you make fixes with a goal in mind, you reach that goal
Whatever it gained in realism, it lost in bias.
Are you going to argue a single fight where a single party member takes 15 damage is enough to stop and take a short rest?
Do you call that credible? credible? Yes. Because it was done by the objective parameters that you set up.
The 10 000 fights model the possibly outcomes of, what donjon defines as a medium or hard challenge, by the alphabetically first monster without special abilities.
In that fight, both fighters spend their action surge, and the BM spends 1 SD. that SD is spend on riposte or on crit damage.
regardless of the result
Someone objective might verywel see a senario where the champion is better:
With their lack of options, champions don't need to worry about saving up their resources, but instead have their ability always up. That makes them better in fights where they are unable to gauge the enemy.
A BM might wrongfully expect a decent fight from a manticore, and save his SD to use them for maximum efficiency in a long fight; accidently dragging out the fight.
Oppositely, a chamption will will mop that manticore up faster, and go defend another partymember.
You, in your bias, see an artifact that must be removed.
credible? Yes. Because it was done by the objective parameters that you set up.
As I said: "The test must be a representative sample of a short-rest's worth of encounters."
How else can you compare a short-rest feature to a permanent one?
As I said: "The test must be a representative sample of a short-rest's worth of encounters."Yeah, you said that ... after the champion got an ever-so-slightly advantage over the BM. You also suggested advocated deadly encounters and metagaming.
Before, the test those 10K iterations represented one fight. In that fight, both fighters spend their action surge, and the BM spends 1 SD. that SD is spend on riposte or on crit damage.
Now that you don't like the result, you move the goalpost.
But hey, go right ahead. A representative sample however can only be reached with a represantive party, not a 2 man party of the same class, but a 4 man party of striker, healer, controller and the champ/BM.
... I wonder how you will stick to that idea.
Zalabim
2017-07-01, 05:59 AM
Manticore barely has time for 1-2 hits before going down. Action surge on the first round is enough to kill it usually. Not a fight a DM would try, at least not without giving the manticore some range to soften up the fighters.
If you give the monster an advantageous starting position, or disadvantageous starting position, that changes the expected difficulty of the encounter up or down, possibly a whole step.
If you add a second manticore to make it a real fight instead of a speed bump, you return to the usual statistically insignficant gain for BM.
One manticore is supposed to be a normal fight. Two manticores is supposed to be deadly. Not speed bumps.
As I said: "The test must be a representative sample of a short-rest's worth of encounters."
How else can you compare a short-rest feature to a permanent one?
You had never said that. You never tested that to begin with. Two level 5 fighters have a daily adventure XP budget of 7000 XP. That means a short rest should be expected after fighting encounters totaling ~ 2333 XP in difficulty. The hardest fight you've tested was 6 Apes (1500 adjusted XP). 4 Apes is 1000 adjusted XP. 1 Mantictore is 1050 adjusted XP. 2 Manticores is 2800 adjusted XP. The encounter thresholds for these two fighters together are Easy 500, Medium 1000, Hard 1500, and Deadly 2200.
The fighters may or may not want a short rest after two medium or one medium and one hard encounter. 2 Manticores at once is straight to a deadly encounter, and they still handle it well. I'd wager that's because the close range constitutes a disadvantage for the manticore, so it's more like 1 easy/2 hard than 1 medium/2 deadly. Long range probably would be a disadvantage for these fighters though, so it'd be 1 hard/2 deadly+. If we're talking about artifacts in the testing process, and not a displayed bias against the champion.
Hold on, hold on.
I'm sorry, what?
Since when making cool fights is the purpose of this exercise?
I thought it was about determining facts, not making stuff look awesome. What kind of chemist would go "I've added copper to our new fuel. It serves no purpose, but the green fire will be sick as hell when we test it"?
Also, I don't know for you, but as a DM I wouldn't make the PCs fight groups of apes of various sizes three times back-to-(silver)back or more. Unless if I was being facetious and my players didn't mind me monkeying around.
A D&D chemist totally would. Possibly a goblin. D&D chemists aren't really well versed in science.
I'd wager that's because the close range constitutes a disadvantage for the manticoreNot really. Just look at the other straight forward CR 3 monsters:
Owlbear: 59 hp, AC 13, init+1 /// +7(10), +7(10) <no ranged attacks>
Nightmare: 68 hp , AC 13, init +2 /// +6 (20=13+7fire) <no ranged attacks>
Manticore: 68 hp, AC 14, init +3 /// +5(6), +5(6), +5(7)
Veteran: 58, AC 17, init +1 /// +5(7), +5(7), +5(6)
With alphabetically, the nightmare being the next monster, yet having even lesser defenses (ac, init) then the manticore, I don't see it surviving longer longer ...
Heck, considering the BM uses his SD for riposting, the single attack nightmare will have even less chance to trigger it. (chances of one 14 in three rolls is higher then one 13 in a single roll)
Zalabim
2017-07-01, 08:12 AM
Not really. Just look at the other straight forward CR 3 monsters:
Owlbear: 59 hp, AC 13, init+1 /// +7(10), +7(10) <no ranged attacks>
Nightmare: 68 hp , AC 13, init +2 /// +6 (20=13+7fire) <no ranged attacks>
Manticore: 68 hp, AC 14, init +3 /// +5(6), +5(6), +5(7)
Veteran: 58, AC 17, init +1 /// +5(7), +5(7), +5(6)
With alphabetically, the nightmare being the next monster, yet having even lesser defenses (ac, init) then the manticore, I don't see it surviving longer longer ...
Heck, considering the BM uses his SD for riposting, the single attack nightmare will have even less chance to trigger it. (chances of one 14 in three rolls is higher then one 13 in a single roll)
That's interesting. It has pretty reasonable defenses for its CR. The owlbear is clearly in the next higher category for offensive CR with its attack bonus. Though the manticore's offense doesn't really seem to be the limiting factor in its performance.
I should admit I have a theory that fights with multiple enemies tend to last longer than fights of equivalent difficulty featuring fewer enemies of higher CR instead.
Now that you don't like the result, you move the goalpost.
The goalpost were never moved, only the limitations of the simulation were.
The whole purpose of this simulation has always been to compare a short-rest feature to a permanent one.
I've had enough of your uncouth behavior. It's time you stop rampaging and get that fact through your thick skull.
JNAProductions
2017-07-01, 11:16 AM
I've had enough of your uncouth behavior. It's time you stop rampaging and get that fact through your thick skull.
Oh, the irony!
One manticore is supposed to be a normal fight. Two manticores is supposed to be deadly. Not speed bumps.
That why I hacked the code to make it 2 consecutive fights.
You had never said that. You never tested that to begin with. Two level 5 fighters have a daily adventure XP budget of 7000 XP. That means a short rest should be expected after fighting encounters totaling ~ 2333 XP in difficulty. The hardest fight you've tested was 6 Apes (1500 adjusted XP). 4 Apes is 1000 adjusted XP.
Making a system that simulate a single fight was hard enough. Making enough fights to simulate consecutive encounters was out of the question.
The initial test ate 25% of the party's health and used a single SD. I felt that was the best I could come up that represented a fair fraction of a short rest. You could string 2 in a row and expect both fighters to be at half-health. Or string 4 in a row and use up all SD/hp.
Then someone asked to see what happened with action surge, amongst other things. Now you cannot assume the second encounter matches the first one. You cannot derive the full short rest result, which is uninteresting.
So I made 3 tests:
- the deadly 2 manticores, easy to do but the wrong approach
- no action surge, a better easy one
- 2 consecutive encounters
I can accept that you didn't see that comparing BM to champion would require extending the results to a full short rest. I didn't explicitely state the obvious even if the initial simulation supported it. That was my mistake.
I will not accept further trial of intent. I've stayed true to my goal and civil in my replies. I will not be goaded by trolls' innuendos.
JNAProductions
2017-07-01, 12:11 PM
I will not accept further trial of intent. I've stayed true to my goal and civil in my replies. I will not be goaded by trolls' innuendos.
No... You haven't. You definitely haven't been civil.
No... You haven't. You definitely haven't been civil.
Is this like that sheep-loving bridge builder: "Ahh, but reply to a single troll"?
Or is it that you found some replies to innuedo?
You happy? Squirm Squirm Squirm.
Or maybe I didn't do enough name calling against constructive criticism?
So, why don't you man up. Stop hiding behind ambiguities and say: "your rant about zealots wasn't civil".
There, you got a personnal uncouth reply!
JNAProductions
2017-07-01, 01:09 PM
You've been rude and intellectually dishonest. When the results didn't please you, you changed the test to make the results fit. The rudeness is just plain on display.
If you were host, you'd admit that, even under your simulation, sometimes the Champion does better, because that's what the results show. They also show that the differences between the two are relatively minor-hot or cold dice can affect the outcomes far more than your subclass choice.
KorvinStarmast
2017-07-01, 02:57 PM
I find the other classes silly and overpowered. Moon Druid, paladin, warlock, mystics, certain cantrips (EB, GFB, BB) IMO tend to ruin the game.
No, they don't. (Then again, Mystics are still UA and play test, so if you let them ruin your game at your table, at least it's in the name of play test, right? :smallcool:) Moon Druid has a power spike in Tier 1, by mid Tier 2 not so much. The people who cry about EB still mystify me. As a warlock, I had to be careful with my resources, since we didn't always get two short rests each day, which made me consider my invocation choices carefully, and enjoy being able to contribute with EB to help the party ... PS EB CAN MISS! Just like any other attack. Nobody at our table has wasted their time on BB nor GFB since they are conditional and fiddly. When someone finally uses them, I may change my view.
I find the majority of charisma based casters are overpowered and multiclassing between them is silly and not worth my time. Paladin/warlock sounds cheesy, but if you can pull it off go for it. WoTC should have stuck to their guns and made Warlock Int based. The "fans" who wanted to keep Warlock Ch based are the problem here, as was WoTC listening to them.
But 5E, unlike, 1st and 2nd addition is allowing you to play want you want to play.
That is nice.
You don't have to play a champion or a paladin, its a choice. Do you want to play a grunt?Fighter works Ranger was so damned hard to qualify for ...
Some players believe in concept vs crunch and optimization, and in 5E you can have both or none. Good point, and it's a nice feature of 5e.
Also, I don't know for you, but as a DM I wouldn't make the PCs fight groups of apes of various sizes three times back-to-(silver)back or more. Unless if I was being facetious and my players didn't mind me monkeying around. We don't have a cleric, and have had to fight zombies for 6 of the last 8 sessions we've run. (there's a reason for this, as zombie apocalypse is a sub theme for the campaign ...) We've learned a few things about zombie fighting. I finally got our EK to use a blunt weapon.
Vogonjeltz
2017-07-01, 03:07 PM
In those 80 rounds (with 2 SR as you stated), the BM would use riposte for 2d6+5+1d8 ~ 16 each, 5 success is enough to cover the pidly 66 damage from extra crits, assuming you manage to miss with the 7 others. And as shown in the first post of the thread, the BM will have spent his whole hp pool, twice.
There goes to show how blissfull the zealots of the Cognitive Dissonance Church are
Interesting theory bid, however this also tethers the Battlemaster to a reactive strategy with situational requirements (melee attack that misses the Battlemaster) which may never come to pass. So, at the cost of any additional utility it's possible they could average more if the right conditions come to pass.
If those conditions don't come to pass, however, then the Battlemaster seriously underperforms.
It's not cognitive dissonance to acknowledge that there are caveats to an idea.
10 is way too high.
Why don't you stat out a Hard encounter and actually round by round game it out where the PCs don't burn short/long rest resources. When that doesn't happen, the weaker characters will get incapacitated extending the duration of the fight. Which the PCs might win (on average), albeit with quite a bit of HP loss.
1 maybe 2 hard per rest, tops, and 10 rounds is a really long and tedious fight. Also, a complete a complete lack of resource expenditure is a bit weird.
The point is that we should have the actual baseline for a fight with zero resource expenditure for when the party actually runs out of them. There's going to be at least a couple of those fights per day.
80 rounds a day is way too high, even by your already high estimates.
I did give a range, 80 would be if there were 8 hard fights.
For level 5, against Apes, 4 is a medium encounter and the game would expect 7 of those encounters a day. 5 is also a medium encounter but the game would only expect 5.6 of those encounters. 6 Apes is a Hard encounter and the game expects only 4.66~ of those encounters a day. The game does not expect a party to face 8 Hard encounters in a day. Maybe they could, but don't bet on it.
We're talking about different things.
I meant the number of encounters, not the number of apes.
The 6 Ape fight in this thread is a Hard encounter and the two fighters using no resources end that fight in about 2 to 6 rounds. Their average is 4 rounds, and that's on the long end because they're facing multiple creatures and using single target attacks. If they were facing fewer, more dangerous creatures, the fight would end faster either way. Rounding up the Hard encounters per day to 5, they fight for an average of 20 rounds per day. 80 is right out.
Challenge Rating is based on the assumption that it's a party of about 4 characters, not 2.
A Hard encounter vs 4 characters of level 5 is between 1500 and 4399 xp threshold.
That's:
1 = A Tyrannosaurus Rex, Frost Giant, Young Bronze/Green dragon, Hydra, or Hezrou
2 = A young white dragon and a White Dragon wyrmling, A Banshee and a Wraith, An Ettin and a Hill Giant, a Weretiger and a Werebear, or a Helmed Horror and a Flesh Golem
3-6 = A Ghost and 5 Specters.
7-10 = 7 Quadrones and 3 Tridrones, 8 Gnolls and 2 Gnoll Pack Lords, A Hobgoblin Captain and 9 Hobgoblins, or A Bugbear Chief 4 Bugbears and 5 Goblins
11-14 = 14 Apes (or Blackbears, Hobgoblins, Orcs, Shadows, Worgs, Giant Wasps, etcetera), A Gnoll Pack Lord and 10 Gnolls (An Orc eye of Gruumsh and 10 Orcs), or 4 Dire Wolves with 3 Worgs and 7 Wolves.
15+ = 21 Zombies, 43 Twig Blights, or 109 Crawling Claws (or an angry mob of 109 Pitchfork wielding Commoners).
It doesn't matter what classes you pick, nobody at level 5 has the action economy to kill 109 Crawling Claws (or rampaging peasants) in under 10 rounds without expending resources. Tops they'd kill like, 80ish assuming they killed one with every attack and no attacks missed. Which is itself statistically impossible given that a 1 always misses.
And that's before we consider that there could be a deadly fight, taking significantly more rounds to hash out. (i.e. someone who didn't heed the warnings above the threshold of an ante-chamber doorway releases a slavering horde of ravenous and uncontrolled zombies and skeletons! (say, 30-40).
DanInStreatham
2017-07-01, 04:30 PM
Trying to keep up with a busy thread so apologies if you've changed the scenario/code and I missed it.
One thing that struck me is that if the apes focus fire one fighter at a time before moving onto the next, and the battlemasters only use sd for riposte or criticals, and both fighters nearly always survive: the one fighter who is never getting hit in most of the simulations is unlikely to use all his sd (since he never gets a chance to riposte and won't crit that often).
So I think the setup is suboptimal for the battlemaster.
Not meant to be a criticism because you've attempted to model a pretty complex situation, and been open with the code used.
Unoriginal
2017-07-01, 06:56 PM
We don't have a cleric, and have had to fight zombies for 6 of the last 8 sessions we've run. (there's a reason for this, as zombie apocalypse is a sub theme for the campaign ...) We've learned a few things about zombie fighting. I finally got our EK to use a blunt weapon.
Maybe, but zombies aren't apes (unless zombie apes), and they're known to attack in big groups that then attracts other zombies when the fight starts.
That's pretty different of "we've been attacked by groups of big monkeys all day"
KorvinStarmast
2017-07-01, 07:34 PM
Maybe, but zombies aren't apes (unless zombie apes), and they're known to attack in big groups that then attracts other zombies when the fight starts. That's exactly what happens. If we had a cleric who could turn undead it would make the ability to "divide and conquer" at least plausible. As is, we hit a few, then as we get mobbed typically withdraw to avoid getting swarmed.
That's pretty different of "we've been attacked by groups of big monkeys all day" Is this a return to the Champion discussion, or a reference to the "hate for the fighter archetypes" thread? :smallbiggrin:
What, the actual f*** !?
In actual play, people don't do average damage. In actual play, in each combat, the swinginess of dice (tyranny of small numbers) is part of play.
While I appreciate the effort you've gone to, the problem of that kind of statistical analysis is that it can't account for the non linear effects of "how the fighter is doing things as integrated with the other party members." While I found the exercise of some interest, I don't find it of any practical value. But thanks for grinding the numbers anyway.
Hairfish
2017-07-01, 08:08 PM
I've stayed true to my goal and civil in my replies.
Proof that the Dunning-Kruger effect also applies to social skills. Tell us some more about how calling anyone who disagrees with you the "Cognitive Dissonance Brigade" is an example of you engaging in civil discourse.
You've been rude and intellectually dishonest. When the results didn't please you, you changed the test to make the results fit. The rudeness is just plain on display.
Right.
The initial test was 2 fighters vs 4 apes, no action surge.
The last single-encounter test was 2 fighters vs 1 manticore, no action surge.
If you say using 100% of your action surge in 1/4 of your short rest can be repeated to represent the full short rest, you are intellectually dishonest.
If you say I meant to measure the value of SD vs improved critical in a single-round battle, you are making a trial of intent.
If you would rather see nefarious deed where none exist and refuse to believe a coherent explanation, you are being rude.
Ask questions, do not throw accusations.
Proof that the Dunning-Kruger effect also applies to social skills. Tell us some more about how calling anyone who disagrees with you the "Cognitive Dissonance Brigade" is an example of you engaging in civil discourse.
There are 2-3 members of this forum (which I will not name) who rehash the same empty fact-free discourse whenever one of the 6-8 people who've actually calculated it mention their result. Never have they shown any numbers to back their claim.
Could you agree with, let's say, mghamster's story about that? Or would you close up and say that the champion can keep going for 100 rounds after the SD are spent?
I am sorry you've felt targetted, but I don't think the hat fits you.
mgshamster
2017-07-01, 10:03 PM
Could you agree with, let's say, mghamster's story about that?
Partially unrelated:
I've recently put all my analysis on excel to keep track of my math, based on previous errors (forum math is a bitch).
My most recent unpublished results show that at level 20, a half-orc great axe champ with GWM performs better than the BM over the course of a standard day. No advantage needed.
But, there are (multiple) reasons I havent published it on the forums yet. I still need to ensure accuracy.
So please, anyone reading this, don't quote this as accurate.
Hairfish
2017-07-01, 10:09 PM
There are 2-3 members of this forum (which I will not name) who rehash the same empty fact-free discourse whenever one of the 6-8 people who've actually calculated it mention their result. Never have they shown any numbers to back their claim.
Could you agree with, let's say, mghamster's story about that? Or would you close up and say that the champion can keep going for 100 rounds after the SD are spent?
I am sorry you've felt targetted, but I don't think the hat fits you.
My position is that (a) you haven't demonstrated a statistically-significant difference between the two subclasses even for the limited models given and - more importantly - (b) the limited models given are terrible metrics by which to gauge performance.
Yes, it's really difficult to model the performance of a particular subclass in a variety of party makeups and encounters. That doesn't mean that these simple models provide useful data... in fact, it means it's highly unlikely that it's even possible to obtain useful data using simple models.
Your critics could bring up the laundry list of particular flaws with the models, but that doesn't overcome the core problem of the models being too simple to reliably translate to real-play situations. Why would it matter that (for example) a level 9 champion gets +2 to initiative over an identically-built BM, that different subclasses wouldn't pick the same equipment or attributes anyway, that BMs trying to keep a pack of apes from the party would use Goading Attack over Riposte, etc, etc, when they don't address the problem that your approach is fractally wrong?
You appear to have already convinced yourself that your models are useful, otherwise you wouldn't have started this thread. What numbers could I possibly present you with, using your simple models, that would convince you that the problem is that simple models aren't going to provide useful numbers here?
Cybren
2017-07-01, 10:11 PM
My position is that (a) you haven't demonstrated a statistically-significant difference between the two subclasses even for the limited models given and - more importantly - (b) the limited models given are terrible metrics by which to gauge performance.
Yes, it's really difficult to model the performance of a particular subclass in a variety of party makeups and encounters. That doesn't mean that these simple models provide useful data... in fact, it means it's highly unlikely that it's even possible to obtain useful data using simple models.
Your critics could bring up the laundry list of particular flaws with the models, but that doesn't overcome the core problem of the models being too simple to reliably translate to real-play situations. Why would it matter that (for example) a level 9 champion gets +2 to initiative over an identically-built BM, that different subclasses wouldn't pick the same equipment or attributes anyway, that BMs trying to keep a pack of apes from the party would use Goading Attack over Riposte, etc, etc, when they don't address the problem that your approach is fractally wrong?
You appear to have already convinced yourself that your models are useful, otherwise you wouldn't have started this thread. What numbers could I possibly present you with, using your simple models, that would convince you that the problem is that simple models aren't going to provide useful numbers here?
Right, these models only serve to show off a particular difference in the exact scenario modeled. They would be very useful if you needed to know which subclass to bring to kill a group of apes as quickly as possible though.
Yes, it's really difficult to model the performance of a particular subclass in a variety of party makeups and encounters. That doesn't mean that these simple models provide useful data... in fact, it means it's highly unlikely that it's even possible to obtain useful data using simple models.
I agree, that's basically what became clear once Oerlaf and qube educated me on confidence interval and all. That made it clear the randomness of dice rolls (and the chunkiness of overkill) was bigger than the extra damage from archetypes by... lets call it a factor of 4.
You seem convinced there are no way to reduce the error bar to something meaningful for archetype or class comparison. I believe you are right and see no reason to explore it further.
Well, I tried something that didn't work. It was a fun toy to build.
Zalabim
2017-07-02, 02:48 AM
Partially unrelated:
I've recently put all my analysis on excel to keep track of my math, based on previous errors (forum math is a bitch).
My most recent unpublished results show that at level 20, a half-orc great axe champ with GWM performs better than the BM over the course of a standard day. No advantage needed.
But, there are (multiple) reasons I havent published it on the forums yet. I still need to ensure accuracy.
So please, anyone reading this, don't quote this as accurate.
I have come up with the basic numbers for that level showing the same relationship, though there's obviously wiggle room based on the rate of getting 'cleave' attacks from kills, the rate of losing 'cleave' attacks due to distance or end of encounters, and the value of 'cleave' attacks themselves based on varying target AC and other defenses.
The initial test ate 25% of the party's health and used a single SD. I felt that was the best I could come up that represented a fair fraction of a short rest. You could string 2 in a row and expect both fighters to be at half-health. Or string 4 in a row and use up all SD/hp.
I'm guessing this 25% is why you think the fight's represent 25% of a short rest. They do not. The easier fights are ~43% of a short rest, and the harder fights are ~64%. This is just based on the DMG guidelines. In terms of actual resources spent, 25% of the group's max HP is about 13.5% of the group's daily HP. That is to say, the two fighters have 5 1d10+2 hit dice, and 44 Max HP, and took an average of 22 damage. They can't do that 12 times. They can do it 7 times then someone is getting knocked out. The expectation is to use only some of your HD on each short rest rather than waiting until 0 HP and using all your HD on one short rest.
Of course, this has also all seemed to ignore second wind.
We're talking about different things.
I meant the number of encounters, not the number of apes.
For level 5, against Apes, 4 is a medium encounter and the game would expect 7 of those encounters a day. 5 is also a medium encounter but the game would only expect 5.6 of those encounters. 6 Apes is a Hard encounter and the game expects only 4.66~ of those encounters a day. The game does not expect a party to face 8 Hard encounters in a day. Maybe they could, but don't bet on it.
Emphasis added. I'm talking about encounters too.
Challenge Rating is based on the assumption that it's a party of about 4 characters, not 2.
A Hard encounter vs 4 characters of level 5 is between 1500 and 4399 xp threshold.
That's:
1 = A Tyrannosaurus Rex, Frost Giant, Young Bronze/Green dragon, Hydra, or Hezrou
2 = A young white dragon and a White Dragon wyrmling, A Banshee and a Wraith, An Ettin and a Hill Giant, a Weretiger and a Werebear, or a Helmed Horror and a Flesh Golem
3-6 = A Ghost and 5 Specters.
7-10 = 7 Quadrones and 3 Tridrones, 8 Gnolls and 2 Gnoll Pack Lords, A Hobgoblin Captain and 9 Hobgoblins, or A Bugbear Chief 4 Bugbears and 5 Goblins
11-14 = 14 Apes (or Blackbears, Hobgoblins, Orcs, Shadows, Worgs, Giant Wasps, etcetera), A Gnoll Pack Lord and 10 Gnolls (An Orc eye of Gruumsh and 10 Orcs), or 4 Dire Wolves with 3 Worgs and 7 Wolves.
15+ = 21 Zombies, 43 Twig Blights, or 109 Crawling Claws (or an angry mob of 109 Pitchfork wielding Commoners).
It doesn't matter what classes you pick, nobody at level 5 has the action economy to kill 109 Crawling Claws (or rampaging peasants) in under 10 rounds without expending resources. Tops they'd kill like, 80ish assuming they killed one with every attack and no attacks missed. Which is itself statistically impossible given that a 1 always misses.
And that's before we consider that there could be a deadly fight, taking significantly more rounds to hash out. (i.e. someone who didn't heed the warnings above the threshold of an ante-chamber doorway releases a slavering horde of ravenous and uncontrolled zombies and skeletons! (say, 30-40).
109 crawling claws wouldn't be one encounter since they couldn't all engage the 4 PCs at the same time, and even if they were and could, I'm talking about the daily adventure XP expectations which say that the harder the encounters are in the day, the fewer encounters can be beaten that day. The 4360XP angry peasant encounter represents 31% of the party's daily XP budget. That's closer to 1/3 of the party's daily encounters than 1/8. If they actually tried to fight this out without using resources, and weren't simply killed by the horde for such bravado, they'd each need to kill about 87 or 88 crawling claws (or peasants) that day, and that may actually take over 80 rounds for a wizard or cleric who refuses to spend spell slots.
Finally, I think the system assumptions include the party using resources in encounters. Just a hunch.
busterswd
2017-07-02, 05:12 AM
I agree, that's basically what became clear once Oerlaf and qube educated me on confidence interval and all. That made it clear the randomness of dice rolls (and the chunkiness of overkill) was bigger than the extra damage from archetypes by... lets call it a factor of 4.
You seem convinced there are no way to reduce the error bar to something meaningful for archetype or class comparison. I believe you are right and see no reason to explore it further.
Well, I tried something that didn't work. It was a fun toy to build.
You used coding to model something you enjoy doing, which is something everyone should try at some point. While the thread's civility as a whole kind of dipped near the end, it was an interesting project.
Although in this case, I would've advocated, given the parameters (fixed number of ripostes, fixed enemy AC, identical defensive stats for PCs), a simple DPR vs. ape AC comparison. It would've taken a fraction of the work and given you a more flexible model to insert other stats, since essentially what you're comparing is killing efficiency and the snowball effect it can have on survivability.
That's the conclusion of your (initial) test - and it's a conclusion highly relevant to D&D gameplay.
It is not some artifact of your test - it's not something that should be removed simply because the values don't suit you.
Also, this is super important and not something you should toss aside because it's disagreeable.
The purpose of simulations, experiments, and tests should be to disprove your hypothesis, not to reinforce your conclusion. If results don't agree with your hypothesis, you don't throw out your results, you throw out your hypothesis, because even if it's an edge case, that means there's something you didn't account for.
I've had enough of your uncouth behavior. It's time you stop rampaging and get that fact through your thick skull.Sorry mate, but that's because it seems to be the only way to get through to you.
You need to get it through your thick skull that results are to be analysed to make a conclusion, opposite to be analysed to make alterations to the test so they reach the conclusions you want. That's being intelectaully dishonest, wether you will accept that or not.
If truely, as you claim, the whole purpose of this simulation has always been to compare a short-rest feature to a permanent one. then you got your answer: Other then all the values not changing lying within a margin of error, your test also indicate that
With their lack of options, champions don't need to worry about saving up their resources, but instead have their ability always up. That makes them better in fights where they are unable to gauge the enemy.
A BM might wrongfully expect a decent fight from a manticore, and save his SD to use them for maximum efficiency in a long fight; accidently dragging out the fight.
Oppositely, a chamption will mop that manticore up faster, and go defend another partymember.
That's the conclusion of your (initial) test - and it's a conclusion highly relevant to D&D gameplay.
It is not some artifact of your test - it's not something that should be removed simply because the values don't suit you.
djreynolds
2017-07-02, 07:15 AM
What I realize is that here on the forum, we could all rock a champion.
We could all find away to make it work, right race, right feats, right skills, right "dips".... regardless of the party we are in
But the reality is it "us" the players who are doing the heavy lifting.
Shield master works for any S&B class, GWM/SS are awesome because bounded accuracy.
I love the fighter class and the champion archetype, I choose them because I like "me" versus the world and I don't like to admit I'm wrong.
I will find away to kill the dragon, regardless of what I'm playing, as I'm sure of everyone on this forum.
I'm not a Mentat, I do not trust logic or statistics, "90% of statistics are made up"
What I want out of the champion, is just more fun with the class. Something more, as the French call a certain "I don't know what!"
My change to the archetype is fun, when a champion scores a crit... the rest of party for that round is blessed. If the champion rolls a 1, the party suffers from the bane spell.
We have tried if someone else in the party scores a crit, the champion can use his reaction to make an attack... no one shows up a champion.
What the champion lacks is flavor.
What I realize is that here on the forum, we could all rock a champion.
...
I do not trust logic or statistics
...
What the champion lacks is flavor.Ironically, as the statistics show no significant difference, they agree with you ...
As for their lack of flavor, a this is a thread about how balanced they are - cold hard numbers and all that - while it might be true, it's not relevant.
A BM might wrongfully expect a decent fight from a manticore, and save his SD to use them for maximum efficiency in a long fight; accidently dragging out the fight.
Oppositely, a chamption will mop that manticore up faster, and go defend another partymember.
That's the conclusion of your (initial) test - and it's a conclusion highly relevant to D&D gameplay.
It is not some artifact of your test - it's not something that should be removed simply because the values don't suit you.
That I can agree with. The BM might not use any archetype feature in a single encounter with this result. They'll still be available in the next fight where they might be used, or not.
This will hold true if there's a string of easy fight to cover the whole short rest. As long as you don't need to use action surge to end it quickly, it's repeatable.
That I can agree with. The BM might not use any archetype feature in a single encounter with this result. They'll still be available in the next fight where they might be used, or not.If there's a next fight, of course. I've never seen anyone reason:
Guys, the wizard has gotten a massive hit, so maybe we should take a short r ... oh wait, I see the BM still got some SD's left. Lets keep moving, shall we?
If there's a next fight, of course. I've never seen anyone reason:
Guys, the wizard has gotten a massive hit, so maybe we should take a short r ... oh wait, I see the BM still got some SD's left. Lets keep moving, shall we?
Yep, all 3 manticore spikes hitting the wizard would do 21 damage, a hefty chunk of its ~32 hp. That can happen once in a while.
Hp is the ultimate resource that defines the length of a day, I would say at most 20 rounds of combat* without needing a short rest to replenish it. You can do much better if you start spending SR/LR resources, but 40 rounds would exceed the 5-8 encounters expectation by a lot.
* {assuming repeated manticore encounters against raw fighter:
- 1st: 1 round with action surge, 10 hp lost,
- 2-6th: 3 rounds without, 15 hp lost,
- both fighters are left with 5 hp or so, a huge risk taken.
And yes, I rounded 19 to 20.}
I've updated the first post and title to reflect the results.
Powered by vBulletin® Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.