KPR comparisons

[Galithar]

I'm looking for a compilation of KPR, or submissions of builds and their KPR. I'm specifically interested in the KPR of high level martials in comparison to high level casters and high level martial/caster multiclasses.

KPR is kills per round. You compare your DPR to the AC and HP of a level appropriate enemy. Linked sources can use their own calculations, but if you'd like to submit anything I suggest keeping to a standard of a creature of CR equal to the build based on the DMG creating a monster suggestions. So a level 20 build goes to the chart and looks for the suggested AC and HP of a CR 20 creature.

KPR = (DPR*(1-.05(Enemy AC - Hit Bonus -1)))/Enemy HP

So if I have a DPR of 45 with a +10 to hit against an enemy of AC 17 and 150 HP my KPR is (45*(1-(.05*(17-10-1)))/150 or .21
Meaning I take out (on average) 21% of an enemies HP per round.

Note: This is not an actual build, I'm just using random numbers to give an example. I'll be posting my own build or two when I have them completed.

[LudicSavant]

That KPR formula is incorrect, since it does not account for factors such as crit chance (which makes a huge difference for certain builds).

[MaxWilson]

When you say "level-appropriate enemies", are you going by Xanathar's tables or DMG tables or what? For four level 12 PCs, a CR 13 Beholder is a Medium fight, but so are 4 CR 3 Yetis and 10 CR 1/2 Hobgoblins. The Yetis + Hobgoblins will result in more KPR since they are sixteen of them, but also a higher probability of TPK because they are actually tougher in most ways and more mobile and have a higher DPR against moderate AC.

Also, KPR is a worse metric than HP loss ratios.

[Zigludo]

Boy, there's just so many variables. Particularly when dealing with high level spell casters things get extremely noodly.

Are we talking about spellcasters that just use all of their slots as quickly as they can to nova, or are we talking about KPR over 3 rounds, or are we talking about KPR over 6 3-round combats with 2 short rests? How many enemies are we fighting at once? If for instance the Wizard gets the opportunity to hit two different targets with Meteor Swarm in one turn his KPR doubles. On that subject how are we treating enemy resistances? What about Legendary Resistance?

[Galithar]

That KPR formula is incorrect, since it does not account for factors such as crit chance (which makes a huge difference for certain builds).

It is correct. Crit chance should be factored into your average DPR, not something you only account for in KPR.

You are much better with the math then I am though, so if you'd like to help out and give a full formula for DPR as well I'd be glad to edit a DPR formula into the original post as well.

I'm not 100% sure I know how to calculate things like increased crit chance and GWF dice re-rolls to be willing to put a formula out for others to judge. :P

[Unoriginal]

A creature of CR X is a Medium encounter of 4 adventurers of lvl X, which means it's estimated to be able survive around three rounds against 4 PCs most of the time. So using said CR X creature for a PC of lvl X is not going to help much.

Also, over how many encounters is the KPR supposed to be calculated? A Nova build might do a lot of damage... by spending a lot of ressources in one fight and then running empty, while something like a pure Champion Fighter would certainly be able to deliver the hurt for longer successions of encounters.

[LudicSavant]

Crit chance should be factored into your average DPR, not something you only account for in KPR.

Yes, factors such as AC, crit chance, and hit chance should be factored into your average DPR.

However, in your formula you appear to be accounting for enemy AC and to-hit bonus separately.

KPR = (DPR*(1-.05(Enemy AC - Hit Bonus -1)))/Enemy HP




You are much better with the math then I am though, so if you'd like to help out and give a full formula for DPR as well I'd be glad to edit a DPR formula into the original post as well.

I'm not 100% sure I know how to calculate things like increased crit chance and GWF dice re-rolls to be willing to put a formula out for others to judge. :P

This should include formulas for all that stuff: https://docs.google.com/document/d/11eTMZPPxWXHY0rQEhK1msO-40BcCGrzArSl4GX4CiJE/edit#heading=h.5qcgsqvtvf8v

[Zigludo]

Yeah, you can't really get an accurate DPR number that factors in crit chance, and then multiply that number by your hit rate, because it'll give you inaccurate results due to math reasons. Weak explanation, but, it's late here.

Just use LudicSavant's DPR calculator in his signature, I'm pretty sure it's got everything you want.

[Galithar]

When you say "level-appropriate enemies", are you going by Xanathar's tables or DMG tables or what? For four level 12 PCs, a CR 13 Beholder is a Medium fight, but so are 4 CR 3 Yetis and 10 CR 1/2 Hobgoblins. The Yetis + Hobgoblins will result in more KPR since they are sixteen of them, but also a higher probability of TPK because they are actually tougher in most ways and more mobile and have a higher DPR against moderate AC.

Also, KPR is a worse metric than HP loss ratios.

KPR is often calculated against a single target. But for casters you can give AoE effects based on the DMGs 'suggested enemies affected' for Theater of Mind combat. Just make a note that it's an AoE KPR and if desired also lost a single target. I don't have the book in front of me but it's based on shape and size of the affect.

As for appropriate CR, that's why I made my own definition. It may not be the best but equal to level. A level 12 build compares to a CR 12 Creature.

Also I don't even know what HP loss ratio is, but KPR is what I'm looking for. Feel free to explain HP loss ratio if you'd like though, I enjoy learning new things! Haha

[Galithar]

Yes, factors such as AC, crit chance, and hit chance should be factored into your average DPR.

However, in your formula you appear to be accounting for enemy AC and to-hit bonus separately.






This should include formulas for all that stuff: https://docs.google.com/document/d/11eTMZPPxWXHY0rQEhK1msO-40BcCGrzArSl4GX4CiJE/edit#heading=h.5qcgsqvtvf8v

See I told you you were better at the maths lol

[Chronos]

Another problem with converting DPR to KPR is overkill. A character with three attacks might be able to kill three weak enemies in a round, but a character with one strong attack can only kill one, even if that one attack does as much damage as the three combined.

[Unoriginal]

Another problem with converting DPR to KPR is overkill. A character with three attacks might be able to kill three weak enemies in a round, but a character with one strong attack can only kill one, even if that one attack does as much damage as the three combined.

Quite true. I don't think Rogues would face well on any KPR challenge, for example.

[MaxWilson]

Also I don't even know what HP loss ratio is, but KPR is what I'm looking for. Feel free to explain HP loss ratio if you'd like though, I enjoy learning new things! Haha

HP loss ratio is

DPR / enemy's DPR

If I can cut my enemy's DPR in half by reducing my own DPR by only 20%, e.g. by casting Blur instead of Divine Favor, I'm coming out ahead, and I'll be able to complete more encounters per day.

Generally the highest-DPR enemies are going to be the low-CR swarms: one CR 9 Nycaloth (5000 adjusted XP) has 123 HP and attacks at +9 for 36 HP of damage per round, but 50 kobolds (also 5000 adjusted XP) have 250 HP and attack at +4 plus advantage for 225 HP of damage per round. So if you base your calculations purely on the assumption that you'll be fighting high-CR enemies, you'll overestimate your chances against low-CR enemies that will stomp you into oblivion. E.g. a Divine Smite paladin would do well against the Nycaloth, maybe could even solo it, but the kobolds would annihilate him.

[Nhorianscum]

HP loss ratio is

DPR / enemy's DPR

If I can cut my enemy's DPR in half by reducing my own DPR by only 20%, e.g. by casting Blur instead of Divine Favor, I'm coming out ahead, and I'll be able to complete more encounters per day.

Generally the highest-DPR enemies are going to be the low-CR swarms: one CR 9 Nycaloth (5000 adjusted XP) has 123 HP and attacks at +9 for 36 HP of damage per round, but 50 kobolds (also 5000 adjusted XP) have 250 HP and attack at +4 plus advantage for 225 HP of damage per round. So if you base your calculations purely on the assumption that you'll be fighting high-CR enemies, you'll overestimate your chances against low-CR enemies that will stomp you into oblivion. E.g. a Divine Smite paladin would do well against the Nycaloth, maybe could even solo it, but the kobolds would annihilate him.

To be fair the freindly neighborhood (horribly optimized) blaster can just blow everything and nuke either option into space in one round because tier 2 blasters are thilly.

--------------

In terms of a previous post claiming rouges are a bit screwered here that's really not an issue with theives/AT's. This kicks in early for AT with magic items but functions for both smoothly in tier 3.(Orzhova AT is a thing I still need to do a writeup on for the eclectic build thread).

That said class dpr is a thing I tend to not take seriously at all. Wand of magic missile exists.

[Chronos]

On the other hand, a spellcaster with area-effect spells might be able to wipe out all 50 of the kobolds in a single round.

[MaxWilson]

To be fair the freindly neighborhood (horribly optimized) blaster can just blow everything and nuke either option into space in one round because tier 2 blasters are thilly.

To be fair, that's only because (and only if) kobolds are stupid. If they spread out throughout a middle-sized cavern, using pack tactics to negate disadvantage from long range, they do less damage but the blaster will have a much tougher time killing more than a few at a time. In loss ratio terms :) assuming Fireball Formation is obviously a terrible strategy for them, and they will only do it if the DM wants them to use a terrible strategy so the blaster can feel awesome.

A middle-of-the-road approach would be for the DM to put 25 of them in a big clump that the blaster can feel awesome about nuking, and scatter the other 25 around the battlefield to get revenge after the nuke goes down.

[Unoriginal]

To be fair the freindly neighborhood (horribly optimized) blaster can just blow everything and nuke either option into space in one round because tier 2 blasters are thilly.


The best blast spell in the game does on average 120 points of damages. That's not enough to one-shot the Nycaloth, and that's using their highest spell slot.

Blasters are great against large groups of weak enemies, but they're certainly not going to be the DPS-masters against one singular opponent.



In terms of a previous post claiming rouges are a bit screwered here that's really not an issue with theives/AT's. This kicks in early for AT with magic items but functions for both smoothly in tier 3.(Orzhova AT is a thing I still need to do a writeup on for the eclectic build thread).

Not sure what you mean by that? Rogues do good damages, but I know few builts that'd allow them to kill the most mooks in a many vs few fight, compared to other classes.


To be fair, that's only because (and only if) kobolds are stupid. If they spread out throughout a middle-sized cavern, using pack tactics to negate disadvantage from long range, they do less damage but the blaster will have a much tougher time killing more than a few at a time. In loss ratio terms :) assuming Fireball Formation is obviously a terrible strategy for them, and they will only do it if the DM wants them to use a terrible strategy so the blaster can feel awesome.

A middle-of-the-road approach would be for the DM to put 25 of them in a big clump that the blaster can feel awesome about nuking, and scatter the other 25 around the battlefield to get revenge after the nuke goes down.

A fair point as well.

Thing is simply comparing combat numbers, be it DPK or others, usually go either of two ways: they're tactic-agnostic and only the numbers are compared even if it means those numbers are only relevant if the characters on both side just stand there at a given range and try to damage each other mindlessly regardless of intellect or temperament, or they assume that the one tactic used works at 100% and always.

[Nhorianscum]

The best blast spell in the game does on average 120 points of damages. That's not enough to one-shot the Nycaloth, and that's using their highest spell slot.

Blasters are great against large groups of weak enemies, but they're certainly not going to be the DPS-masters against one singular opponent.



Not sure what you mean by that? Rogues do good damages, but I know few builts that'd allow them to kill the most mooks in a many vs few fight, compared to other classes.

Optimized Sray cast at 5th level will hit 123 in a round under hexcurse+hex on a highish roll (this has become my default blasters do ok damage example, sorry). With empowered and a wand of the war mage + advantage from flyby this isn't an unlikely roll.

Theif picks up a magic stave or wand of fireballs with UMD? AT jacks silly things like spirit guardians with a rav background and dashes. Either one can rip through anklebiters real quick. Same dealy gets us haste + BB.

[Kurt Kurageous]

The metric is flawed. The name is flawed. The concept is interesting.

I propose a metric that is a ratio of damage done to hit points of build. Sorta like bench pressing your weight.

A build with 30 HP doing 20 DPR has a ratio of 2/3 for a basis of comparison. This allows comparison to monsters as well as builds.

The name? I leave that to the OP.

[MaxWilson]

Thing is simply comparing combat numbers, be it DPK or others, usually go either of two ways: they're tactic-agnostic and only the numbers are compared even if it means those numbers are only relevant if the characters on both side just stand there at a given range and try to damage each other mindlessly regardless of intellect or temperament, or they assume that the one tactic used works at 100% and always.

Sophisticated analysis is hard, but it wouldn't be too much to ask for a 50/50 mix: assume half the fights are a best-case scenario for a given build, and the other half the fights are a worst-case scenario (or something close to it).


Optimized Sray cast at 5th level will hit 123 in a round under hexcurse+hex on a high roll (this has become my default blasters do ok damage example, sorry)

Unfortunately, Nycaloths have AC 18 and are resistant to fire damage. Even with +9 to hit and Hex, you're looking at avg 7.att 18 9 3d6/2+d6 (https://shiningsword.blob.core.windows.net/public/v0.3/index.html#battle) = 38.76 DPR against the Nycaloth with that combo, not 123. (If the Nycaloth has pre-cast Darkness to give you disadvantage, you're looking at only 21.57 DPR.)

[Nhorianscum]

Sophisticated analysis is hard, but it wouldn't be too much to ask for a 50/50 mix: assume half the fights are a best-case scenario for a given build, and the other half the fights are a worst-case scenario (or something close to it).



Unfortunately, Nycaloths have AC 18 and are resistant to fire damage. Even with +9 to hit and Hex, you're looking at avg 7.att 18 9 3d6/2+d6 (https://shiningsword.blob.core.windows.net/public/v0.3/index.html#battle) = 38.76 DPR against the Nycaloth with that combo, not 123. (If the Nycaloth has pre-cast Darkness to give you disadvantage, you're looking at only 21.57 DPR.)

Elemental adept is a thing we take when using fire?

We're implying warlock levels. Who have an infamous invocation at second level. Something something see in darkness.

[Man_Over_Game]

I'm not entirely sure why KPR has to be a thing that's separate from DPR.

The only variable that'd change between characters is DPR. More than likely, someone's hit chance isn't going to differ from one character to another, as Proficiency is the same for everyone per level, and their relevant modifiers are only going to differ by about 1-2 points.

You're effectively asking for the highest DPR builds, then modifying the DPR by -5% to +5% based on how high their primary stat is.

[Unoriginal]

Other things that would have to be taken into account would also be the character's survivability (especially if they don't win the initiative) and ressource management.

Being the best at killing things fast doesn't help if the build you're compared with kills half as fast as you do but can stay on the battlefield three times longer than you.

[MaxWilson]

Elemental adept is a thing we take when using fire?

We're implying warlock levels. Who have an infamous invocation at second level. Something something see in darkness.

...I see. You've got a Hexblade warlock who for some reason took Elemental Adept (fire) and Devil's Sight, even though he's planning on using his concentration on Hex instead. Let's suppose that you rolled an 18 Cha so you can still get +9 to hit as before. Congratulations, you can do 80.50 damage in the second round against the Nycaloth.

I'm underwhelmed. That's a lot of investment build and time investment for a middling return. Even odds someone else in the party will have already dealt with it by then.


Other things that would have to be taken into account would also be the character's survivability (especially if they don't win the initiative) and ressource management.

Being the best at killing things fast doesn't help if the build you're compared with kills half as fast as you do but can stay on the battlefield three times longer than you.

That's why HP loss ratios are more interesting than DPR, and why e.g. Careful Web is often a better spell than Haste.

[Nhorianscum]

...I see. You've got a Hexblade warlock who for some reason took Elemental Adept (fire) and Devil's Sight, even though he's planning on using his concentration on Hex instead. Let's suppose that you rolled an 18 Cha so you can still get +9 to hit as before. Congratulations, you can do 80.50 damage in the second round against the Nycaloth.

I'm underwhelmed. That's a lot of investment build and time investment for a middling return. Even odds someone else in the party will have already dealt with it.

Since that's pretty much GWM+PAM+Action surge damage off of a BM using presicion attack I'm not sure why this isn't qualified as "yeah ok, that's pretty decent" damage.

Was assuming sorlock built for blasting tbh pure lock isn't getting Sray unless fiend.. Since that's sorta their thing. So if we slap empower + adept rerolls + Critrange on that... It's an extreme example but yeah we "can" roast it R2 in a white room.

A more realistic scenario in a fight worth burning this much gas on is hexcurse+ fireball R1 to clear flack, SRay R2 to blick big bad. Again though, this is a silly whiteroom damage D measuring contest.

(We're only ever taking 2-3 blast spells all game long even on a dedicated blaster. The only good to-hit is s fire fsr)

[MaxWilson]

Since that's pretty much GWM+PAM+Action surge damage off of a BM using presicion attack I'm not sure why this isn't qualified as "yeah ok, that's pretty decent" damage.

Because it takes two rounds to pull off and costs two 5th level spell slots (Hex and Scorching Ray) and a feat (Elemental Adept) and your limited-use, close-range Hexblade's Curse. As I said, even odds someone else in the party will have already dealt with the Nycaloth by then. For example, if there's a melee Enchanter (e.g. Forge 1/Enchanter 2+), he's got better than 60% chance of locking it up with his at-will Hypnotic Gaze on the first round, at which point it's solved: you can grapple it and toss on a Net and maybe some manacles for restraining and have everybody Ready attacks to go off as soon as the Enchanter moves away from it; and if it survives that the Enchanter still has his action for e.g. Tasha's Uncontrollable Laughter. Beating a CR 9 creature with an at-will feature on round 1 plus maybe a 1st level spell on round 2 is better than beating it with all of your spell slots and a feat and a short-rest feature on round 2.

(This works particularly well if there's a Shepherd Druid in the party with a swarm of conjured animals. 8 Giant Poisonous Snakes or Dimetrodons attacking the Nycaloth at advantage for it being restrained is a lot of damage.)

[Nhorianscum]

Because it takes two rounds to pull off and costs two 5th level spell slots (Hex and Scorching Ray) and a feat (Elemental Adept) and your limited-use, close-range Hexblade's Curse. As I said, even odds someone else in the party will have already dealt with the Nycaloth by then. For example, if there's a melee Enchanter (e.g. Forge 1/Enchanter 2+), he's got better than 60% chance of locking it up with his at-will Hypnotic Gaze on the first round, at which point it's solved: you can grapple it and toss on a Net and maybe some manacles for restraining and have everybody Ready attacks to go off as soon as the Enchanter moves away from it; and if it survives that the Enchanter still has his action for e.g. Tasha's Uncontrollable Laughter. Beating a CR 9 creature with an at-will feature on round 1 is better than beating it with all of your spell slots and a feat and a short-rest feature on round 2.

...

I am confused by this.

[MaxWilson]

I am confused by this.

What is unclear?

I'm giving an example of how I see beefy mid-CR creatures like Nycaloths and young dragons quickly and cheaply dealt with at the table, if they engage in straightforward melee, to explain why I'm unimpressed by an round 2, 80-HP damage nova, as an aside from discussing why KPR isn't a very interesting metric compared to loss ratios.

Does everyone use tactics like these? Nope. Some people just hit it with an axe, or blast it with a spell, or cast Grasping Vine on it, and the Nycaloth will do much better against those players. Your Scorching Ray nova is far from the worst strategy I've seen, but it doesn't come close to fulfilling the promise that "To be fair the freindly neighborhood (horribly optimized) blaster can just blow everything and nuke either option into space in one round because tier 2 blasters are thilly." It can't nuke either option into space in one round, kobolds or Nycaloth.

BTW if you're now assuming a Sorlock with 2 Warlock levels, and 5th level spell slots, you're talking at least a Warlock 2/Sorc 9+, which means you're no longer talking about Tier 2 blasters anyway.

[Galithar]

Okay, first off this thread was made because I was looking for KPR comparisons so I appreciate no one responding on topic. :P

Second KPR is a metric that can show how well a build can do at different levels. It shows the comparison of damage to an enemy death and this changes as you level. Builds with increasing DPR change at a different rate than their KPR changes.

HP loss ratio sounds like an even less useful metric. It shows only if I deal more damage, but doesn't show anything valuable, like if I live long enough. If I deal twice as much damage as the enemy, but he has three times as much health as me HP loss ratio only tells me I deal more damage. KPR simply tells me, on average, how long it will take me to kill him.

It's not supposed to be a perfect metric that does everything and gives a perfect snapshot of a build. It does what it's meant to do slightly better then just DPR because it lets me know the average time it takes me to kill the kinds of things the DMG expects me to fight. No it's not perfect and everything goes out the window as soon as you leave the white room.

[Man_Over_Game]

Okay, first off this thread was made because I was looking for KPR comparisons so I appreciate no one responding on topic. :P

Second KPR is a metric that can show how well a build can do at different levels. It shows the comparison of damage to an enemy death and this changes as you level. Builds with increasing DPR change at a different rate than their KPR changes.

HP loss ratio sounds like an even less useful metric. It shows only if I deal more damage, but doesn't show anything valuable, like if I live long enough. If I deal twice as much damage as the enemy, but he has three times as much health as me HP loss ratio only tells me I deal more damage. KPR simply tells me, on average, how long it will take me to kill him.

It's not supposed to be a perfect metric that does everything and gives a perfect snapshot of a build. It does what it's meant to do slightly better then just DPR because it lets me know the average time it takes me to kill the kinds of things the DMG expects me to fight. No it's not perfect and everything goes out the window as soon as you leave the white room.

A question, then. What is the difference between gathering high DPR builds vs gathering high KPR builds?

[Unoriginal]

A question, then. What is the difference between gathering high DPR builds vs gathering high KPR builds?

Kill Per Rounds depends on factors that make basically everyone but nova builds outclassed. Provided it's only calculated when they can nova.

[Man_Over_Game]

Kill Per Rounds depends on factors that make basically everyone but nova builds outclassed. Provided it's only calculated when they can nova.

I guess I meant that in regards to the formula he's using. Or, if he wants us to come up with builds that use information outside of the formula, how does he want us to determine the effectiveness of a build?

[MaxWilson]

Okay, first off this thread was made because I was looking for KPR comparisons so I appreciate no one responding on topic. :P

Second KPR is a metric that can show how well a build can do at different levels. It shows the comparison of damage to an enemy death and this changes as you level. Builds with increasing DPR change at a different rate than their KPR changes.

That just means the fights are getting longer, not harder, against that monster distribution. Not useful info per se.


HP loss ratio sounds like an even less useful metric. It shows only if I deal more damage, but doesn't show anything valuable, like if I live long enough. If I deal twice as much damage as the enemy, but he has three times as much health as me HP loss ratio only tells me I deal more damage. KPR simply tells me, on average, how long it will take me to kill him.

You misunderstand. HP loss ratios will show you exactly this information that you're looking for. If you're looking at two builds, and one of them does 12 HP of damage for every HP you lose (as a group) against monster distribution XYZ, while the other build does 6 HP of damage for every one you lose, then you know that the first build will keep you alive longer and win you more fights against that monster distribution. KPR will tell you how short the fight is, but loss ratios will tell you which build is better at winning the fight.

I admit that loss ratios are harder to generalize because you need to make assumptions about environment, monster tactics and PC tactics. In the extreme case, when you've got a squishy AC 10 bard who loves to charge onto the frontline, the only way to optimize your loss ratio is to nova your damage as hard as possible to kill monsters before they can kill the idiot bard. But if the bard behaves intelligently and/or wears heavy armor, you have more options.

[Nhorianscum]

What is unclear?

I'm giving an example of how I see beefy mid-CR creatures like Nycaloths and young dragons quickly and cheaply dealt with at the table, if they engage in straightforward melee, to explain why I'm unimpressed by an round 2, 80-HP damage nova, as an aside from discussing why KPR isn't a very interesting metric compared to loss ratios.

Does everyone use tactics like these? Nope. Some people just hit it with an axe, or blast it with a spell, or cast Grasping Vine on it, and the Nycaloth will do much better against those players. Your Scorching Ray nova is far from the worst strategy I've seen, but it doesn't come close to fulfilling the promise that "To be fair the freindly neighborhood (horribly optimized) blaster can just blow everything and nuke either option into space in one round because tier 2 blasters are thilly." It can't nuke either option into space in one round, kobolds or Nycaloth.

BTW if you're now assuming a Sorlock with 2 Warlock levels, and 5th level spell slots, you're talking at least a Warlock 2/Sorc 9+, which means you're no longer talking about Tier 2 blasters anyway.

1: 7 sorc points = 5th level slot. 7/2 isn't my prefered split at this point but as a poor mans proof of concept it works.

2: Great, this is a KPR thread? Snozbrerries are snozzberries.

3: The key words are "horribly optimized" and "can" this is a KPR thread where we swing digs about who can nova who down in burst based on % chance and turns taken. (Also somone does math eventually and I get to enjoy that)

4: Again.Nothing stops a click boom blasty guy from running a full support/BC suite. Not many good booms to chose from. Inflcting the deadstatus is sometimes useful.

5: Sorlock was used because it's common as sin and I like Sorc. There are harder novas and harder sorc Novas but this one is basic and easy to crunch with a huge damage varience based on rolls, investment of slots/time, and magic items that is good for a KPR thread.

Obviously real play differ's from whiteroom and "I seduce cthulu" becomes an if not THE most optimal strat. Again snozzberries are snozzberries

[Galithar]

A question, then. What is the difference between gathering high DPR builds vs gathering high KPR builds?

Honestly, not a hell of a lot. I'm asking because another poster (in another thread) was very adamant about the importance of KPR and I'm just doing some 'Independent research' on the subject.

Which requires me to see some KPR stats on different builds. The difference between DPR and KPR is simply the ratio to average monster health at the end. So if someone were to just give me some more DPR information I could do the extra number crunching so long as they give me an Average DPR assuming all hits. Otherwise the DPR must also be calculated against the proper AC.

I was mostly hoping someone would be able to link me somewhere where this had already been done for a number of builds to quickly gather information without me having to piece together builds and then calculate KPRs myself.

I think both metrics, KPR and DPR, are near useless the second you step out of the white room, though more useful than HP loss ratio because that has far too many variables and provides less useful information to me. If I lose 6 HP per one I deal on average, but end the combat at the beginning of round 2 does your HP loss ratio account for that? No. If I lose 12 HP per one, but 3 times per day I can Nova half the encounter away and reduce it to an effective 6 per 1 can the HP loss ratio calculate that? It tries to do more then it's capable. Everytime you add more variables your data gets less useful because the margin for error is greater. By this I think DPR is a better metric then KPR which is better then HP loss ratio. Each step adds more variables and becomes less reliable. HP loss ratio is simply comparing my DPR to the enemies, effectively doubling the variables and ruining my margin of error. All of them are useless to be in actual play because they account only for optimal conditions with average rolls. I don't know about any of you, but that's not what actually happenes at my tables. Lol

[thereaper]

That other poster would be me.

DPR by itself is a meaningless metric. It only matters in comparison to the target. 15 dpr is amazing at first level, but terrible at 18th level.

The reason why this matters is that according to the KPR chart I've seen (https://docs.google.com/spreadsheets/d/1d-9xDdath8kX_v7Rpts9JFIJwIG3X0-dDUtfax14NT0/edit#gid=2035285798), KPR generally decreases as you level. In other words, monsters tend to scale faster than dpr. So, any damage-focused build (which is usually martials, but can also represent certain spellcasters as well) will scale more slowly than a build that is tailored for other things (or a build that tries to be balanced). This leads almost inevitably to the conclusion that builds designed for damage to the exclusion of all else will contribute less to a party than more balanced builds past a certain point.

This is also the reason why you see people claiming spellcasters are still better than martials this edition (since most spellcaster power is in the form of buffs, debuffs, utility, etc that aren't affected by this, or are affected less; and martials are generally KPR specialists).

If that particular KPR chart is flawed, or if there are builds nowadays that can increase their KPR over time, I would personally consider it a Very Big DealTM, and would be very pleasantly surprised.

[MaxWilson]

HP loss ratio is simply comparing my DPR to the enemies, effectively doubling the variables and ruining my margin of error.

Improving that ratio is the whole point of every decision you make in combat. The goal is to kill monsters without getting dead.

Yes, it's harder to compute, but that's also why it's more accurate than KPR/DPR at predicting who will win the fight. KPR/DPR leaves out half of the important variables!

[Man_Over_Game]

That other poster would be me.

DPR by itself is a meaningless metric. It only matters in comparison to the target. 15 dpr is amazing at first level, but terrible at 18th level.

The reason why this matters is that according to the KPR chart I've seen (https://docs.google.com/spreadsheets/d/1d-9xDdath8kX_v7Rpts9JFIJwIG3X0-dDUtfax14NT0/edit#gid=2035285798), KPR generally decreases as you level. In other words, monsters tend to scale faster than dpr. So, any damage-focused build (which is usually martials, but can also represent certain spellcasters as well) will scale more slowly than a build that is tailored for other things (or a build that tries to be balanced). This leads almost inevitably to the conclusion that builds designed for damage to the exclusion of all else will contribute less to a party than more balanced builds past a certain point.

This is also the reason why you see people claiming spellcasters are still better than martials this edition (since most spellcaster power is in the form of buffs, debuffs, utility, etc that aren't affected by this, or are affected less; and martials are generally KPR specialists).

If that particular KPR chart is flawed, or if there are builds nowadays that can increase their KPR over time, I would personally consider it a Very Big DealTM, and would be very pleasantly surprised.

No, it's probably very accurate.

A Rogue is considered a fairly high damage option. This SHOULD be so, as the Rogue's main contribution to combat is dealing damage, unlike a Fighter (who can mitigate hits and be an obstacle to ranged attackers) or a Cleric (who can heal allies and weaken enemies).

A Rogue's Sneak Attack gains 1d6 damage every 2 levels. Or 1.75 damage every level.

The average amount of HP a character should gain per level, with a +1 modifier, is about 6. This is taken by averaging the various hit-die sizes (1d6-1d12), and slightly upscaling it due to the Tough feat being an option (but there's no method of having LESS HP per level).

Health scales about 3x faster than damage.

Magic occasionally breaks this trend, but max level spell slots become more sparse as you level up. Upgrading in spell slot levels always starts you off with 2 slots of your max level, until you hit level 7 (when every new upgrade after that starts with a single max level spell slot).

[LudicSavant]

See I told you you were better at the maths lol

:smallredface:


If that particular KPR chart is flawed

Kryx's chart appears to be using a great deal of eyeballing and assumptions (like "duration of Concentration effects is constant and doesn't depend on your build" or "nova damage doesn't shorten encounters" or "polearms have a flat 50% chance of getting an OA" or "spellcasters don't target saves opportunistically"), and doesn't use particularly optimized builds for its numbers (as Kryx himself mentions on the first page).

Actually checking his math is challenging because of the way his sheet is entangled with everything else on the sheet, so just trying to figure out the formula he's using to arrive at a single number is... a time-consuming process of reverse engineering. Seriously it would be really helpful if I could just see the formula for any single conclusion laid out on a single line. But it looks to me like he's not accounting for some variables, such as optimal use of maneuvers in his Battle Master writeups (which should basically be "use Precision Attack if your initial roll misses, use a different maneuver if you hit").

[Galithar]

Improving that ratio is the whole point of every decision you make in combat. The goal is to kill monsters without getting dead.

Yes, it's harder to compute, but that's also why it's more accurate than KPR/DPR at predicting who will win the fight. KPR/DPR leaves out half of the important variables!

It's not more accurate because it's so much harder to compute. The fewer variables you have the more accurate your metric will be.

They don't leave out 'half the important variables' they calculate the important variables for determining what they do. Which is 'How much hurt can I deal?'

The amount of hurt you deal compared to the enemy is extremely useful, but also extremely swingy. Things like AC don't swing as much as enemy damage (or even health which is why I prefer DPR, but am looking at KPR because it was brought up in a good discussion on another thread)

For example a Bone Devil Polearm averages 64 damage in a round. Another CR 12 the Warlord (in Volo's) averages 24.

Their ACs are 1 point apart (19 and 18 respectively)
Their HPs are 87 points apart (142 and 229 respectively)

DPR against each will remain similar. KPR is better against the Bone Devil Pole Arm and the HP loss ratio would be better against the Warlord.

They all do different things, but DPR is the most consistent and therefore in my eyes most valuable metric.

[LudicSavant]

Here's an example of what sort of damage a fighter can do when they're working together with their party and optimizing a bit.

Calculations are against an AC 19 target (DMG average for level 20).



Samurai 20 (Sharpshooter/Crossbow Expert/Elven Accuracy/Max Dex):
Fighting Spirit / Rapid Strike: 85.4 DPR
FS / RS / Action Surge: 160.3 DPR
FS / RS / AS / Strength Before Death: 320.6 DPR

Note that the Samurai in particular scales incredibly well with buffs / teamwork due to AS, RS and SBD. Let me give you an idea of how well:
FS/RS/Elemental Weapon (1 hour buff, cast at 7th level): 133.45 DPR.
FS/RS/Action Surge + Elemental Weapon: 248.1 DPR
FS/RS/Action Surge + Elemental Weapon + Holy Weapon (they stack, but two different party members need to do it): 333.8 DPR

FS/RS/AS/EW/SBD: 496.2 DPR
FS/RS/AS/EW/HW/SBD: 667.6 DPR
FS/RS/AS/EW/HW/SBD/Haste: 744.7 DPR

So uh, yeah. Cast a buff on your local Samurai today. It's good value for your spell slot/Concentration. (I seriously feel like accounting for teamwork is overlooked WAY too often in character optimization threads. The Samurai scales really well with teamwork due to their features basically acting as a force multiplier for every die that gets added to them).

(Note: I'm usually the one playing the buffer here, so I often calculate who in the party is best to buff ^^;;)

If you want to convert any of these numbers to KPR you just factor in the enemy's max HP.

Edit:
More Fighter info. Here's an Anydice program I wrote to calculate BM vs Samurai nova damage at level 8.

Both builds use point buy.
Samurai is an elf (any elf) with Elven Accuracy (+1 Dex), +2 Dex, and Sharpshooter as their ASIs. Total 20 Dex.
Battle Master is a Dex race (any non-variant dex race) with Sharpshooter, Crossbow Expert, and +2 Dex as their ASIs. Total 18-19 Dex.

Samurai is using Fighting Spirit for triple Advantage. BM is using a maneuver on every attack (Precision Attack if their initial roll misses, a damaging maneuver if they hit). Both are using Action Surge. Target AC is 16 (DMG average for level 8).

https://anydice.com/program/14e93

AC: 16
PROFICIENCY: 5

function: attack DMG:d crit CRIT_DMG:d penalty PENALTY:n rider RIDER:n stat STAT:n superiority SUP:n roll ROLL:n{
MODIFIER: (STAT-10)/2
DMG_WITH_BONUS: DMG + MODIFIER + RIDER

if ROLL = 1 {
result: 0
}
if ROLL = 20 {
result: DMG_WITH_BONUS + CRIT_DMG
}
if ROLL + MODIFIER + PROFICIENCY - PENALTY >= AC {
result: DMG_WITH_BONUS + SUP
}

if ROLL + SUP + MODIFIER + PROFICIENCY - PENALTY >= AC {
result: DMG_WITH_BONUS
}
result: 0
}


output
[attack 1d6 crit 1d6+2d8 penalty 5 rider 10 stat 18 superiority 1d8 roll 1d20]
+ [attack 1d6 crit 1d6+2d8 penalty 5 rider 10 stat 18 superiority 1d8 roll 1d20]
+ [attack 1d6 crit 1d6+2d8 penalty 5 rider 10 stat 18 superiority 1d8 roll 1d20]
+ [attack 1d6 crit 1d6+2d8 penalty 5 rider 10 stat 18 superiority 1d8 roll 1d20]
+[attack 1d6 crit 1d6 penalty 5 rider 10 stat 18 superiority 0 roll 1d20]
named "BM"

output
[attack 1d8 crit 1d8 penalty 5 rider 10 stat 20 superiority 0 roll [highest 1 of 3d20]]
+
[attack 1d8 crit 1d8 penalty 5 rider 10 stat 20 superiority 0 roll [highest 1 of 3d20]]
+
[attack 1d8 crit 1d8 penalty 5 rider 10 stat 20 superiority 0 roll [highest 1 of 3d20]]
+
[attack 1d8 crit 1d8 penalty 5 rider 10 stat 20 superiority 0 roll [highest 1 of 3d20]]
named "Samurai"

Lvl8 CBE Battle Master: 65.00 DPR
Lvl8 Elf Samurai: 70.82 DPR

For the KPR just divide by some specific enemy's hit points (not sure which specific one you'd want to use; the DMG doesn't actually give a flat value for HP by CR, and other monster features often cut into that hit point value).

[MaxWilson]

It's not more accurate because it's so much harder to compute. The fewer variables you have the more accurate your metric will be.

Spherical cows are not real cows.


For example a Bone Devil Polearm averages 64 damage in a round. Another CR 12 the Warlord (in Volo's) averages 24.

Their ACs are 1 point apart (19 and 18 respectively)
Their HPs are 87 points apart (142 and 229 respectively)

DPR against each will remain similar. KPR is better against the Bone Devil Pole Arm and the HP loss ratio would be better against the Warlord.

The task is to find the build, or the action in combat, that has the best HP loss ratio against the expected monster distribution (e.g. one Warlord and one Bone Devil), not to find the monster which has the worst HP loss ratio against a fixed PC. It's not like you can change which monster you're facing.

There's no important difference, tactically, between doubling your DPR and halving the monster's DPR. Either way you end the fight with the same number of HP left. But if you can e.g. cut the Bone Devil's DPR by 50% by casting Wrathful Smite or Protection From Evil, so that you take only 30 HP before you kill it instead of 60 HP, that is usually better than a Divine Smite for 3d8 (14) HP with that same spell slot, which has only a low probability of saving you maybe 15 HP of damage in a fight against a solo Bone Devil.

KPR is useless for telling you what actions or builds are best at winning fights.

[Galithar]

Spherical cows are not real cows.



The task is to find the build, or the action in combat, that has the best HP loss ratio against the expected monster distribution (e.g. one Warlord and one Bone Devil), not to find the monster which has the worst HP loss ratio against a fixed PC. It's not like you can change which monster you're facing.

There's no important difference, tactically, between doubling your DPR and halving the monster's DPR. Either way you end the fight with the same number of HP left. But if you can e.g. cut the Bone Devil's DPR by 50% by casting Wrathful Smite or Protection From Evil, so that you take only 30 HP before you kill it instead of 60 HP, that is usually better than a Divine Smite for 3d8 (14) HP with that same spell slot, which has only a low probability of saving you maybe 15 HP of damage in a fight against a solo Bone Devil.

KPR is useless for telling you what actions or builds are best at winning fights.

First your numbers are faulty there. Disadvantage is not a 50% DPR reduction, it could be more it could be less. Also using a spell slot for a smite and saying you only lose the damage of the smite is inaccurate. You also lose the damage from the full attack action. Action economy is the single most important thing in 5e. Second your HP loss ratio is comparing to specific creatures in this instance. Getting an advantage against one does nothing for the other. Protection from Evil will help in one and not the other. This is a situational variable that you should not be trying to calculate in white room scenarios. What can give you an idea in a white room is how much damage can I deal? Since AC rarely changes drastically at any given CR (there are exceptions, but they are less significant then HP or enemy DPR) you can calculate a few situations and get s better picture. DPR at advantage, DPR straight, and DPR at Disadvantage and you've covered most scenarios. You can calculate these things in a white room. Planning to adjust for an ability or type of enemy isn't practical because you'd end up with too much information to be meaningful.

If I can make a build that does 30, 20, 6 (Adv, straight, disadv) I can know that if I'm going to be attacking at disadvantage I should be looking for another action because my Disadvantage DPR is lacking.

It's not a perfect metric but it is generalized enough to be impactful. Knowing HP loss ratio against a Bone Devil doesn't help me to know how effective I am against a Warlord because it involves tactics. Tactics are not white room friendly. DPR is simply my damage potential and then I use that information when planning my combat tactics.

I get why you like HP loss ratio, but it fluctuates too much to be relevant in a general situation. DPR relatively stays constant. KPR is similarly dependant on the enemies HP. DPR only cares about AC, advantage and disadvantage. Which the other two metrics also care about, but add their own variance on top of the existing variant.

HP loss ratio is superior only in judging the outcome of specific scenarios because it takes specific variables to calculate.

[MaxWilson]

First your numbers are faulty there. Disadvantage is not a 50% DPR reduction, it could be more it could be less. *snip*

Did you know that e.g. stands for "exempli gratia," ("for example")? My math isn't off--I was giving a fairly typical example based on experience. Additional detail:

If I'm a Paladin with AC 19 (plate armor plus defense style), I can normally expect to take 25.73 DPR from a polearm variant Bone Devil, counting crits. (+8 for 2d12+4, +8 for 2d8+4+5d6). Imposing disadvantage cuts that DPR to 11.97, which in this case is just over a 50% reduction in damage taken. If my AC were 21 from using plate armor + shield, it would be closer to a 63% reduction.

Sheesh.


It's not a perfect metric but it is generalized enough to be impactful. Knowing HP loss ratio against a Bone Devil doesn't help me to know how effective I am against a Warlord because it involves tactics. Tactics are not white room friendly. DPR is simply my damage potential and then I use that information when planning my combat tactics.

I get why you like HP loss ratio, but it fluctuates too much to be relevant in a general situation. DPR relatively stays constant. KPR is similarly dependant on the enemies HP. DPR only cares about AC, advantage and disadvantage. Which the other two metrics also care about, but add their own variance on top of the existing variant.

HP loss ratio is superior only in judging the outcome of specific scenarios because it takes specific variables to calculate.

Evaluating the KPR of two different tactics is useless. Evaluating the loss ratios of two different tactics tells you which one to use against that enemy.

I'll quit beating the dead horse now.

[Galithar]

Did you know that e.g. stands for "exempli gratia," ("for example")? My math isn't off--I was giving a fairly typical example based on experience. Additional detail:

If I'm a Paladin with AC 19 (plate armor plus defense style), I can normally expect to take 25.73 DPR from a polearm variant Bone Devil, counting crits. (+8 for 2d12+4, +8 for 2d8+4+5d6). Imposing disadvantage cuts that DPR to 11.97, which in this case is just over a 50% reduction in damage taken. If my AC were 21 from using plate armor + shield, it would be closer to a 63% reduction.

Sheesh.



Evaluating the KPR of two different tactics is useless. Evaluating the loss ratios of two different tactics tells you which one to use against that enemy.

I'll quit beating the dead horse now.

I'm not comparing KPR of tactics. I'm not comparing DPR of tactics. I'm using DPR as a base to know the damage capabilities of a build. That's all white room is good for. I don't look at specific enemies with DPR and only brought them up because it becomes relevant to HP loss ratio.

You're trying to claim a metric is better at something that I'm not asking for. Yes in specific circumstances where you can calculate the variables it gives a great picture. That's useless in a white room... You don't have a specific enemy. You have how much damage can you deal? HP loss ratio REQUIRES details that generally aren't available. You can make a build that is fantastic against undead and make it's HP loss ratio amazing against them. Then you put it against a humanoid and it's making defenses crumble. The DPR doesn't change much... (Again remember AC is the least in flux stat of monsters within a CR).

[MaxWilson]

I'm not comparing KPR of tactics. I'm not comparing DPR of tactics. I'm using DPR as a base to know the damage capabilities of a build. That's all white room is good for. I don't look at specific enemies with DPR and only brought them up because it becomes relevant to HP loss ratio.

You're trying to claim a metric is better at something that I'm not asking for. Yes in specific circumstances where you can calculate the variables it gives a great picture. That's useless in a white room... You don't have a specific enemy. You have how much damage can you deal? HP loss ratio REQUIRES details that generally aren't available. You can make a build that is fantastic against undead and make it's HP loss ratio amazing against them. Then you put it against a humanoid and it's making defenses crumble. The DPR doesn't change much... (Again remember AC is the least in flux stat of monsters within a CR).

You're objecting to the idea of at least trying to make a useful metric, e.g. by computing HP loss ratios against two disparate canonical encounters at each level, because you say it's too hard to do. This reminds me of a story:

A policeman sees a drunk man searching for something under a streetlight and asks what the drunk has lost. He says he lost his keys and they both look under the streetlight together. After a few minutes the policeman asks if he is sure he lost them here, and the drunk replies, no, that he lost them in the park. The policeman asks why he is searching here, and the drunk replies, “this is where the light is.”

Related story:

Milk production at a dairy farm was low, so the farmer wrote to the local university, asking for help from academia. A multidisciplinary team of professors was assembled, headed by a theoretical physicist, and two weeks of intensive on-site investigation took place. The scholars then returned to the university, notebooks crammed with data, where the task of writing the report was left to the team leader. Shortly thereafter the physicist returned to the farm, saying to the farmer, "I have the solution, but it works only in the case of spherical cows in a vacuum".

I've had this argument with Kryx more times than I can count. A metric which is easy to compute but doesn't reflect actual play is useless. It's just a spherical cow.

[Galithar]

You're objecting to the idea of at least trying to make a useful metric, e.g. by computing HP loss ratios against two disparate canonical encounters at each level, because you say it's too hard to do. This reminds me of a story:

A policeman sees a drunk man searching for something under a streetlight and asks what the drunk has lost. He says he lost his keys and they both look under the streetlight together. After a few minutes the policeman asks if he is sure he lost them here, and the drunk replies, no, that he lost them in the park. The policeman asks why he is searching here, and the drunk replies, “this is where the light is.”

Related story:

Milk production at a dairy farm was low, so the farmer wrote to the local university, asking for help from academia. A multidisciplinary team of professors was assembled, headed by a theoretical physicist, and two weeks of intensive on-site investigation took place. The scholars then returned to the university, notebooks crammed with data, where the task of writing the report was left to the team leader. Shortly thereafter the physicist returned to the farm, saying to the farmer, "I have the solution, but it works only in the case of spherical cows in a vacuum".

I've had this argument with Kryx more times than I can count. A metric which is easy to compute but doesn't reflect actual play is useless. It's just a spherical cow.

A metric that measures only a specific circumstance doesn't affect general play. Your metric is only better when given SPECIFICS. Tell all the snarky anecdotes you want, it doesn't make your metric any better in a generally applied metric. DPR doesn't do it all, the only difference is I'm not claiming it does.

[djreynolds]

Okay, first off this thread was made because I was looking for KPR comparisons so I appreciate no one responding on topic. :P

Second KPR is a metric that can show how well a build can do at different levels. It shows the comparison of damage to an enemy death and this changes as you level. Builds with increasing DPR change at a different rate than their KPR changes.

HP loss ratio sounds like an even less useful metric. It shows only if I deal more damage, but doesn't show anything valuable, like if I live long enough. If I deal twice as much damage as the enemy, but he has three times as much health as me HP loss ratio only tells me I deal more damage. KPR simply tells me, on average, how long it will take me to kill him.

It's not supposed to be a perfect metric that does everything and gives a perfect snapshot of a build. It does what it's meant to do slightly better then just DPR because it lets me know the average time it takes me to kill the kinds of things the DMG expects me to fight. No it's not perfect and everything goes out the window as soon as you leave the white room.

1, this is not a terrible idea

2, A while back I tried to come up with WAR (warriors above replacement), like baseball metrics to see how a standard party of four, consisting of a Champion fighter, Thief, Life Cleric and Evoker Wizard would fair versus another competing party or how changing out say a fighter with a paladin or barbarian, or rogue with a bard would make a difference

3, the idea was to include average dice rolls for "to hit" and damage and critical hits and including 3 short rests a day and skill coverage and you could say that your new party was better than the above "average party".

You could say that a party with a paladin, druid, bard and sorcerer could compete or best the "average party" in a set of tasks during a campaign

4, its ends up being just a considerable amount of work to accomplish, and feat selection also changes dynamics... I didn't even included multiclassing

[MaxWilson]

A metric that measures only a specific circumstance doesn't affect general play. Your metric is only better when given SPECIFICS. Tell all the snarky anecdotes you want, it doesn't make your metric any better in a generally applied metric. DPR doesn't do it all, the only difference is I'm not claiming it does.

I apologize if that came across as snarky. It was intended to illustrate a point, not to put you down personally.

[djreynolds]

I tried the W.A.R. awhile back.

A cleric may spam bless from 1st to 4th level while the battlemaster goes to town with SS and GWM... but at 5th level the cleric says hey I like spirit guardians

Its really tough to gauge team play, especially through multiple tiers of levels

[JackPhoenix]

You know what? I'll throw few things your way. You've mentioned level 20... so I'll give you randomly generated adventuring day, using kobold fight club (http://kobold.club/fight/#/encounter-builder). Going for medium/hard encounters while staying within recommended budged for adventuring day. 2 short rest, take them anytime you want between encounters. Enviroment is not taken into account.

First set is for party of 4 level 20 characters. 160k adjusted xp daily budget.
Vampire, 5x air elemental: hard, 38k xp
Chain devil, 2x cyclops, 3 sea hags in coven: medium, 23 600 xp
Ancient white dragon: medium, 25 000 xp
Beholder in lair, vampire warrior: hard, 36 750 xp
Green abishai (MToF), ice devil: hard, 36 750 xp
160 100 xp total, so almost exactly within the daily budget.

Second set is for one level 20 character. 40k adjusted xp daily budget
Hezrou: medium, 5 850 xp
2x bearded devil, 3x guard drake (VGtM), merregon (MToF): hard, 9 625 xp
Flesh golem, 2x yuan-ti mind whisperer (VGtM): hard, 10k xp
8x fire newt warlock of Imix (VGtM): easy, 4 800 xp
2x hobgoblin devastator (VGtM), 4x peryton: hard, 10k xp
40 275 xp total, again, almost exactly the daily budget.

Should be enough variety, and it shows what level 20 characters are supposed to be dealing with. In the first example, there would be other 3 characters in normal circumstances, which would change things due to mutual support and various synergies, but whatever, you aren't supposed to solo it, just show what your build can do against such opponents.

[OverLordOcelot]

HP loss ratio is

DPR / enemy's DPR

If I can cut my enemy's DPR in half by reducing my own DPR by only 20%, e.g. by casting Blur instead of Divine Favor, I'm coming out ahead, and I'll be able to complete more encounters per day.

How do you account for healing in this model, especially noncombat healing? For example, if you're using healing spirit, lots of people reducing becomes irrelevant - if the most damage anyone takes is 60, then you burn one third level slot between combats to heal everyone to full, and reducing damage for anyone who takes less than the most damage becomes irrelevant. And since the OP mentioned looking at a level 20 build, how does this analysis work with a level 20 moon druid, who can keep refreshing a 100+ hp pool every round? Do you treat enemy damage done as zero if it's within what the druid can refresh on a bonus action, in which case your HP loss ratio tends to infinity, or do something else to account for it?

I'm extremely skeptical of any simple model handling higher level play, especially level 20, because in practice it's pretty much never the steady grind that models like, and instead has a lot of novas and saveable spell effects.

[MaxWilson]

How do you account for healing in this model, especially noncombat healing? For example, if you're using healing spirit, lots of people reducing becomes irrelevant - if the most damage anyone takes is 60, then you burn one third level slot between combats to heal everyone to full, and reducing damage for anyone who takes less than the most damage becomes irrelevant. And since the OP mentioned looking at a level 20 build, how does this analysis work with a level 20 moon druid, who can keep refreshing a 100+ hp pool every round? Do you treat enemy damage done as zero if it's within what the druid can refresh on a bonus action, in which case your HP loss ratio tends to infinity, or do something else to account for it?

I'm extremely skeptical of any simple model handling higher level play, especially level 20, because in practice it's pretty much never the steady grind that models like, and instead has a lot of novas and saveable spell effects.

I think about postcombat healing as something which mitigates HP loss, but only if you survive. In an uberdeadly encounter you burn whatever resources you have to, to survive, even if it's normally very inefficient (Divine Smite). But if you're already winning then you try to win as cheaply as possible, e.g. taking four rounds to kill a bad guy via grapple/prone instead of nova-ing to kill him in two, and then just healing the extra two rounds of damage via Aura of Vitality or Healing Spirit.

You still need insurance against threats that turn out to be more deadly than you thought, but PCs usually have plenty of panic buttons: consumable magic items, limited-use defensive abilities, Lucky/Portent dice, etc. Save those for when they matter, don't just blow them to save 20 HP in a fight you're going to win by a mile anyway. (Unless you're doing it for RP reasons. I end up overusing Shield works just because damage HURTS and I don't like taking it, even if taking it and then healing after with precast Goodberries would be more efficient.)


You know what? I'll throw few things your way. You've mentioned level 20... so I'll give you randomly generated adventuring day, using kobold fight club (http://kobold.club/fight/#/encounter-builder). Going for medium/hard encounters while staying within recommended budged for adventuring day. 2 short rest, take them anytime you want between encounters. Enviroment is not taken into account.

*snip*

...show what your build can do against such opponents.

I don't know who this was addressed to but it sounds like fun. I'll give it a go tomorrow morning if my other hobbies don't get in the way.

[JackPhoenix]

I don't know who this was addressed to but it sounds like fun. I'll give it a go tomorrow morning if my other hobbies don't get in the way.

To nobody in particular. People are arguing about KPR or DPR, but that depends on opposition. I presented variety of enemies that the characters should be expected to face at level 20, made it full adventuring day (even if sticking to medium and hard encounters means the day is pretty easy overall), so it's not biased towards nova build who can burn every resource on single opponent, and even made versions for both party and solo play. It's still a white room scenario, but at least there are now some set expectations.

[MaxWilson]

To nobody in particular. People are arguing about KPR or DPR, but that depends on opposition. I presented variety of enemies that the characters should be expected to face at level 20, made it full adventuring day (even if sticking to medium and hard encounters means the day is pretty easy overall), so it's not biased towards nova build who can burn every resource on single opponent, and even made versions for both party and solo play. It's still a white room scenario, but at least there are now some set expectations.

The most complicated thing about the encounters is the geometry and tactics, because that affects things like the effectiveness of AoE crowd control spells from the PCs. For my analysis I'm planning to assume that all of the fights take place in a 20 yard x 20 yard x 20 yard cube, that the beholder is a special variant that can use eye rays even against targets it cannot see (so heavy obscurement will not trivialize the fight), and that the monsters and PCs both use simple, metagamed tactics such as attacking the squishiest target possible every round and never risking opportunity attacks. Hag Coven will concentrate on spells like upcast Hold Person and Counterspell, because they're simple to model.

[Zuras]

I don’t see how you can construct a singular KPR number that would allow you to compare different characters without being grossly misleading.

At the very least, you need to look at things by encounter, rather than round, and have splits by encounter type.

For example, let’s separate different types of level appropriate encounters. Since we’re trying to create a number for moneyball style analysis, let’s call it the MLB split, for Minion/Lieutenant/Boss monsters.

Minions are expected to be encountered in numbers. Lieutenants are at or slightly below par CR, the party should encounter at least 2 but no more than the total PC count of them. Bosses are above party level CR and likely have legendary actions and resistances.

If you look at things on a by-encounter basis, you can do some additional calculations to gauge the effectiveness of non-damage spells. If we assume 3 rounds of combat, per the CR calculations guidelines, then a spell that incapacitated an enemy for a round is worth 1/3 of a kill, letting us calculate the value of spells like Hold Person and Command, and theoretically letting us place a value on stuff that is obviously valuable but hard to pin down in a damage-only discussion, like spell selections that target multiple saves and abilities that help maintain concentration.

You still have to guesstimate some stuff, like the number of creatures you can catch in an AoE, since obviously Fireball should hit more enemies than Thunderwave under average conditions, but that would be a good starting point.

If your methodology is valid, you should get results indicating Wizards have a MLB Kills Per encounter of (for example) 8/1/0.3, while a Paladin has a 3/1.5/0.75, indicating what we already know, that the Paladin can nova hard against single targets while the wizard excels at clearing out mooks.

[LudicSavant]

A question, then. What is the difference between gathering high DPR builds vs gathering high KPR builds?

There isn't one, because KPR and DPR are directly proportional. KPR is just DPR/Monster HP. As such the highest DPR build and the highest KPR build will be the same build, and vice versa.