Guidelines for avoiding common pitfalls in theory crafting

[Fwiffo86]

I propose a set of "guidelines" for accounting how well classes perform in a given day. Something we can bounce back to to say by Encounter X, on average this amount of resources can be reasonably assumed to have been used. This should give some assistance to avoid the wizards always have their spells, pallys always have their smites, etc.

Using a separate chart for 6 encounters, 7 encounters, and 8 encounters respectively. Concurrently, we set it to midrange difficulty for continuity.

Anyone else have ideas to contribute?

[EvilAnagram]

You'd have to control for certain things. Magical resistance, damage resistance, and immunity to spells/damage are important factors. These are massive difficulty modifiers when only considering single classes.

[Fwiffo86]

I was thinking average expenditure. Yes, you are correct about a myriad of situational modifiers and what not. I'm just looking to do a baseline.

Basically...

Level 20 Wizard had X spell slots. 1xX 2xX, etc. Divided by encounter totals for day for 6/7/8 encounters.

Level 20 Paladin has X spells slots. (as above). Average Smite usage could be say... 1+ per encounter which eats into available slots, etc.

And so forth for each class. Nothing other than assuming a somewhat even expenditure of resources per encounter is all I'm looking for.

Resources to account for:

Fighter - 2nd Wind, Action Surge, Spell slots, Superiority Dice, etc.

Bard - Spell slots, Bardic inspiration, etc.

Wiz - Spell slots, Arcane recovery, etc.

Rogue - Spell slots (AT based)

Druid - Wildshape uses, Spell slots

Warlock - Spell slots, Pact boons

Sorc - spell slots, metamagic points

Monk - Ki usage

Barbarian - Rage uses

And so forth.

[Renvir]

I think a great starting basis for this is to look at DMG pages 82-84. Generally, a second level character of any class by themselves should be able to handle 6 medium difficulty encounters in a day. Medium encounters for that character being fights that have an XP budget of 100. During the day that character would expect to take 2 short rests and finish with a long rest.

When breaking it down by class I think the categories of factors to consider are HP, short rest abilities, long rest abilities, at-will abilities, and permanent effects.
(The assumption I would make is that items, magical and otherwise, should scale with level and encounter difficulty and their effects be held constant. Obviously, a necklace of fireballs could one-shot an encounter but it would also be useless if you were fighting monsters immune to fire.)

These five categories come together to inform how long a character can keep going before they need to stop.
-Obviously, HP needs to be greater than 0 or everything comes to a halt. Otherwise a character can be just as dangerous with 1 HP or 100 HP.
-How much a character relies on short rest abilities to stay relevant can really change the number of encounters. Especially, if the GM won't allow you to constantly take breathers in between fights. (This includes warlock spells, Action Surge, Ki points, etc.)
-Once you use a long rest ability you are out of luck until you get some rest. This is a big part of caster classes being relevant. The higher level a caster is the longer they may stick around.
-Classes with a lot of at-will abilities (I'm thinking extra attack, cunning action, some invocations) really only need to worry about HP to keep going. Classes without much in the way of at-wills, or weak at-wills, will have to be more careful about other ability expenditure as fights progress.
-Permanent effects are things like what EvilAnagram pointed out. I would also include things like AC and Saving Throws here as well since they rarely change day to day.

I've loved GMing 5e because I think the classes were built with these things in mind and situations where one person needs to rest while everyone else wants to keep going have been rare. I hope this proves helpful and that I didn't totally miss the point of your question.

[Fwiffo86]

I completely concur. So, how do we throw that into a table?

[Renvir]

Off the top of my head I would have a table that goes Class, HP, Short Rest, Long Rest, At-Will, Permanent. On the next line I would start with Barbarian and underneath the other headings I would give a Low, Medium, or High dependent rating (or some other rating system). As an example I might say Wizard, HP=Low, Short Rest=Low, Long Rest=High, At-Will=Medium, Permanent=Low/Medium. Those ratings aren't set in stone and their meanings are pretty abstract and need refinement.

Once I do that I would probably add a few columns new based on a number of encounters in a day. Having 1 through 8 seems excessive but isn't a bad place to start. These columns would have a different rating along the lines of Fine, Relevant, and Useless. To determine that rating I would look at where classes got Low, Medium, and High ratings. A class with High HP, Permanent, or At-Will ratings will be Fine for the first several encounters and then drop off to Relevant the longer they go. Classes with a High Short Rest rating would be Fine for 1 or 2 encounters and then start dropping off. By encounter 4 they might be Useless. A High Long Rest rating would be Fine for 1 encounter and then the drop off would depend on how many Long rest abilities they have. A level 1 wizard won't have nearly the staying power of a level 20 wizard.

The tricky part is balancing those things together, but this is a start. Doing this by level or tier of play would come in as well. There's no point in comparing level 5 fighters and level 18 warlocks.

[Fwiffo86]

Off the top of my head I would have a table that goes Class, HP, Short Rest, Long Rest, At-Will, Permanent. On the next line I would start with Barbarian and underneath the other headings I would give a Low, Medium, or High dependent rating (or some other rating system). As an example I might say Wizard, HP=Low, Short Rest=Low, Long Rest=High, At-Will=Medium, Permanent=Low/Medium. Those ratings aren't set in stone and their meanings are pretty abstract and need refinement.

Once I do that I would probably add a few columns new based on a number of encounters in a day. Having 1 through 8 seems excessive but isn't a bad place to start. These columns would have a different rating along the lines of Fine, Relevant, and Useless. To determine that rating I would look at where classes got Low, Medium, and High ratings. A class with High HP, Permanent, or At-Will ratings will be Fine for the first several encounters and then drop off to Relevant the longer they go. Classes with a High Short Rest rating would be Fine for 1 or 2 encounters and then start dropping off. By encounter 4 they might be Useless. A High Long Rest rating would be Fine for 1 encounter and then the drop off would depend on how many Long rest abilities they have. A level 1 wizard won't have nearly the staying power of a level 20 wizard.

The tricky part is balancing those things together, but this is a start. Doing this by level or tier of play would come in as well. There's no point in comparing level 5 fighters and level 18 warlocks.

I was thinking listing resources available by class, dividing up evenly between total encounters of the day. This gives a baseline of how much resources can be assumed to be used per encounter. Figure at level 20 first. As most arguments stem from high powered play.

Don't rank anything. Just simply put on the table. Ignore rest mechanics as they are determined at whim of DM. State that the table does not account for rest mechanics in the beginning.

Level 20 Warrior has X Superiority Dice / 8 = x.x dice per encounter spent.

[Renvir]

That would certainly be a more thorough approach to determining what actions could be undertaken per encounter. As an example we could look at a level 20 Champion Fighter (because they are easily the most straight forward class) over 6 encounters:

Fighting Style: Always active
Second Wind: 1/6
Action Surge: 2/6
Extra Attack: Always active
Indomitable: 3/6
Improved Critical: Always active
Remarkable Athlete: Always active
Survivor: Always active

How does that work for you?

[Tanarii]

Using a separate chart for 6 encounters, 7 encounters, and 8 encounters respectively. Concurrently, we set it to midrange difficulty for continuity.that's too many encounters. With an equally mixed number of Medium and Hard Encounters, a party should be able to hand 5 encounters before needing a Long Rest. (Theoretically)

[Renvir]

6 is a lot but is certainly in the realm of possibility. If all you do is medium encounters then 6 is at or near your max based on WotC math. Throwing in hard or deadly encounters would decrease it but a couple easy encounters could push it above that. Without any short rests you'd have a hard time going past 3 for most any class.

I think part of this exercise is to see just how far a class could push themselves before needing to rest. Anecdotally, Warlocks and Monks wouldn't be nearly as useful after 2 medium to hard combat encounters due to being out of spells and ki points respectively. But a Champion Fighter doesn't really see much of a drop in output until their HP hits 0.

[Ruslan]

The biggest pitfall I see in theorycrafting is the "level 20" pitfall. Basically, a certain "build" is evaluated based on how it functions on level 20, and on level 20 only. There is no consideration of how it plays on the way there. This is not to say that some builds can't be great on all levels (incl. level 20), just that levels 1-19 are the most-often-forgotten stepchild of theory :smallwink:

[Renvir]

For the sake of balancing all class levels but without evaluating each class 20 times I would probably just do it by tier of play. Something like levels 3, 6, 9, 13, and 20. From my own experience campaigns don't seem to hit the higher levels often but throwing in 16 or 17 wouldn't hurt either.

[krugaan]

I mentioned in another thread, but it would be nice to have a chart of a few things that are plotted against level:

Max Burst Damage
Max non-burst damage (consuming no per/rest resources) -
In-Combat Utility
OOC Utility

then it might be easier to see how each class fares, although multiclass builds would be more difficult

[Ruslan]

Another pitfall is over-reliance on damage analysis. There are a lot of other things in a game (I mean, even the DMG mentions Combat as only one of the three pillars, along with Exploration and Interaction).

From personal experience, in a typical D&D session, you're in combat between 1/2 and 1/4 of the time. Rest of the time, you're doing something else. And even when in combat, damage is by far not the only thing that matters. For example, the Open Hand Monk's ability to inflict status conditions is invaluable - even if difficult to quantify in a number. The Mobile feat, as another example, doesn't get the attention is deserves, because it doesn't increase DPR. But, it lets you get away from enemies, and forces them to waste valuable time re-engaging you, and that's gotta be worth something.

Skills and out-of-combat class features don't get enough analysis. Possibly because people don't know how to analyze them, so they fall back on damage. That's simple arithmetics, so most people know how to do it. But when a critical mass of people does so, it creates the false impression that damage is the key to class balance.

If you want to achieve something new and valuable, do more non-damage analysis. DPR was done to death.

[DanyBallon]

It may not be simple to put in a chart, but could we account that some classes are less useful a certain level?

Let me explain, I've always been puzzled by people claiming that wizards were gods in 3.P
Yes they are really powerful at high level, and can overshadow martial classes, but if played beyond the 5 minute adventuring day, they spend most of their early level doing nothing and be a sitting duck most of the relying on their comrades until they become powerhouse.

It was even worse when they had only a d4 for HP :smallbiggrin:

[krugaan]

If you want to achieve something new and valuable, do more non-damage analysis. DPR was done to death.

That's a good idea, in general, except there's really not a lot to go on there, numerically. That's probably why DPR is done all the time, because spreadsheets.

You'd have to set a reasonable system for rating things like OOC utility and such, and then everyone would argue about the ratings, and then everyone would argue about the relative rankings in the ratings...

Is there a way to objectify stuff like that?

[Renvir]

Multi-classing would increase the number of permutations by so much it wouldn't be worth it. Spells would make some of those burst damage numbers insane. Depending on how you handle a Fireball spells radius on a square grid (Diagonals are fun!) a level 5 Wizard could theoretically deal 48 fire damage to 65+ targets resulting in an output of 3120 total damage. My guess is you meant single target burst damage to avoid that silliness.

[krugaan]

Multi-classing would increase the number of permutations by so much it wouldn't be worth it. Spells would make some of those burst damage numbers insane. Depending on how you handle a Fireball spells radius on a square grid (Diagonals are fun!) a level 5 Wizard could theoretically deal 48 fire damage to 65+ targets resulting in an output of 3120 total damage. My guess is you meant single target burst damage to avoid that silliness.

Yeah, in terms of BBEG killing.

It's pretty much a given that casters will trounce everyone else in the AOE department.

[Renvir]

That's a good idea, in general, except there's really not a lot to go on there, numerically. That's probably why DPR is done all the time, because spreadsheets.

You'd have to set a reasonable system for rating things like OOC utility and such, and then everyone would argue about the ratings, and then everyone would argue about the relative rankings in the ratings...

Is there a way to objectify stuff like that?

To really do that you'd have to capture every roll in and out of combat, quantify levels of success and failure, and analyze those outcomes based on all possible outcomes. I enjoy breaking things down to their tiniest parts and putting it all back together but my friends would kick me to the curb if I ever tried making this happen.

[Ruslan]

You can quantify utility in feats or half-feats. That's one idea. For example, a Rogue, compared to a Fighter, gets +2 skills and +1 tool. Coincidentally, proficiency in 3 skills-and-or-tools is exactly worth one feat (Skilled). Expertise in two skills is also approximately worth a feat (in fact I saw a homebrew feat that does just that. It was fairly balanced, as far as homebrew goes).

So you could say the Rogue is two feats ahead of the Fighter for noncombat utility.

[Tanarii]

6 is a lot but is certainly in the realm of possibility. If all you do is medium encounters then 6 is at or near your max based on WotC math. Throwing in hard or deadly encounters would decrease it but a couple easy encounters could push it above that. Yes, 6 is about average for all Medium Encounters. Maybe a bit more. I brought it up because the OP seemed to be assuming 6-8, which is a number given in the DMG text for a mix of medium and hard encounters, that doesn't match the charts.


Without any short rests you'd have a hard time going past 3 for most any class.

I think part of this exercise is to see just how far a class could push themselves before needing to rest. Anecdotally, Warlocks and Monks wouldn't be nearly as useful after 2 medium to hard combat encounters due to being out of spells and ki points respectively. But a Champion Fighter doesn't really see much of a drop in output until their HP hits 0.That didn't seem like part of the exercise at all, based on the OP. The exercise seemed to be what resources a party could expect to have expended across a standard adventuring day. Not for a short-rest-less grind. Purpose being for discussions of inter-class balance across a standard adventuring day.

Kryxx already did something similar for on the expected DPR output side, although IIRC it primarily focused on martials and consistent damage output classes. He had to make assumptions about expected length of encounters based on difficulty, distribution of encounter difficulty, and several other things in the process.

[Kevingway]

Don't assume that your DM will allow you to use optional rules/variants/homebrew/etc. if planning to implement your theory crafting into an actual game. I always like to start with a base of "strictly normal rules," which includes no feats or multiclassing, and work up from there. Learning each base class in and out from 1-20 makes for interesting multiclass interactions when you come to that part, if desired.

[Ruslan]

My experience is that overwhelming majority of DMs allow feats, and most allow multiclassing. In fact, the only time I saw a DM enforce a basic game, was in a game for very new players, who aren't aware of internet optimization anyway.

It is, IMO, safe to assume that the target audience for your guide will have access to most common optional features of the game, such as feats and MC.

[krugaan]

To really do that you'd have to capture every roll in and out of combat, quantify levels of success and failure, and analyze those outcomes based on all possible outcomes. I enjoy breaking things down to their tiniest parts and putting it all back together but my friends would kick me to the curb if I ever tried making this happen.

Sure, but you could generalize to a good extent, or break it down by category and sort of ... average it or something.

OOC utility:
- social skills - bard to barbarian
- general skills - bard to fighter
- spells/abilities - wizard to fighter

Combat Utility:
- crowd control - from wizard to rogue
- mobility - monk/rogue to paladin
- healing/buffing/mitigation - cleric to almost everyone else

the DPS comparisons could be just numbers.

[krugaan]

You can quantify utility in feats or half-feats. That's one idea. For example, a Rogue, compared to a Fighter, gets +2 skills and +1 tool. Coincidentally, proficiency in 3 skills-and-or-tools is exactly worth one feat (Skilled). Expertise in two skills is also approximately worth a feat (in fact I saw a homebrew feat that does just that. It was fairly balanced, as far as homebrew goes).

So you could say the Rogue is two feats ahead of the Fighter for noncombat utility.

Yeah that's one way. Expertise could count as +1 to +1.5 proficiencies per skill. Jack of all trades would be worth ... expertise x num skills / 2.

Think we should start a new thread for this?

[Renvir]

I think a new thread would be good. We've gotten a bit off topic.

[PeteNutButter]

Judging by the title, the OP seems to be saying, or at least implying that NOVA builds are pitfalls in theory crafting, since they won't have resources for the whole day.

I for one, find that any party could do very well with at least one NOVA character, who participates little throughout the medium encounters but goes... well, NOVA when the party is in fear of actually losing a fight. He is the party's "oh ****" button.

I applaud the attempt to categorize all this, if it is a little useless, too many variables IMO.

Also, even the released adventures by WotC don't have much 6-8 encounters per day. I've played them all (working on Strahd), from multiple DMs, and it rarely occurs that there are that many fights. Only a handful of times were there more than a few fights (3-4) per adventuring day, with many days having only one or two.

I suppose its possible that all the DMs ignored the books, since I haven't read them, just played them. Someone who's read them can comment better on that, but if my sample space (admittedly too small to be scientific ~5) is any indication of how the game is played as a whole it means the DMG is a poor reference for an actual adventuring day.

[cZak]

Pretty cool thought theory-ing

Maybe start at 5th level.
Establish the basics of comparison when resources are less might make it easier.
Then expand the base for additional resources. Might help id missing or overvalued ratings.

[Fwiffo86]

Don't assume that your DM will allow you to use optional rules/variants/homebrew/etc. if planning to implement your theory crafting into an actual game. I always like to start with a base of "strictly normal rules," which includes no feats or multiclassing, and work up from there. Learning each base class in and out from 1-20 makes for interesting multiclass interactions when you come to that part, if desired.

This. This here.

The point of this is to chart "somewhat" reliably a count of resource expenditure. Nothing more. We aren't tracking damage. Build the foundation first. Then see what we can add to it.

Again....

No feats. There is no reason to calculate these based on the premise they are "optional". Anything that is optional is not/will not be charted. We are attempting to chart only that which is absolutely 100% guaranteed to be in every game/every table, no questions asked.

No max/min damage calculations. Since most of these require feat calculations, or some other extrapolation we haven't even begun to chart yet, I see no reason to spend energy debating this.

All we care about is resource expenditure. We don't even care what those resource expenditures are for. Its not part of the process.

Decide how many encounters of "midrange" difficulty (as an average between easy-hard) and divide resources. Once we have that accomplished, we can move on to adding other things based on the first set of numbers.

As for Lvl 20 - I choose this not because I have any aspirations to chart each level, but because it will be the "end" for all classes, regardless of game.

[Tanarii]

Decide how many encounters of "midrange" difficulty (as an average between easy-hard) and divide resources. Once we have that accomplished, we can move on to adding other things based on the first set of numbers.the math makes this approximately five, with two short rests.