PDA

View Full Version : D&D 5e/Next How much on-command burst should a class have?



PhoenixPhyre
2017-10-05, 08:22 PM
So I'm working on balancing a homebrew base class (the Seeker (http://homebrewery.naturalcrit.com/share/B12upq1_-)), and I'm trying to do as much as possible before actually building a whole slew of encounters and running a sample party through those. Because that seems like work :smallamused:

I put together a spreadsheet with approximate DPR and sustainable burst DPR (assumptions follow) for the following classes and archetypes (not including my Seeker for this post):


* No feats
* No magic items
* Except the paladin, all Basic Rules characters.
* 6x fights per day, 2 short rests. 2 encounters, short rest, 2 encounters, short rest, 2 encounters, long rest
* Fights last an average of 4 rounds. Any number would work and only change things slightly.
* Sustainable DPR: Try to use burst options equally across all fights. No more than once per turn.
* Fighting creatures mainly of CR = Level - 2 (or so, the absolute numbers change if the CR changes but not the relative numbers, at least by much)
* Calculated at each level where proficiency changes: 1, 5, 9, 13, and 17
* Max attack stat as top priority.

* Mountain Dwarf Champion Fighter with a greataxe and GWF style
* Wood Elf Champion Fighter with a longbow and Archery FS
* Wood Elf Champion Fighter with two shortswords and TWF style
* Wood Elf Rogue (any) with two shortswords. 100% uptime on sneak attack.
* Wood Elf Rogue (any) with a light crossbow. 100% uptime on sneak attack.
* Mountain Dwarf Paladin (no archetype specified for simplicity) with a greataxe and GWF style. Assumption: uses all 4th and lower slots to smite.


What I found was interesting (to me at least). I present the amortized Damage Per Round (including sustainable burst), expressed as a percentage of the GWF Fighter's damage.



Archetype
Level 1
Level 5
Level 9
Level 13
Level 17


GWF Fighter
100
100
100
100
100


TWF Fighter
128
94
99
89
87


Archer Fighter
85
90
86
86
86


TWF Rogue
133
81
107
88
95


Archery Rogue
109
72
99
83
90


GWF Paladin
100
99
156
170
183



Commentary:

*Archery is consistently behind for these archetypes. Both steady-state and burst damage suffer
*TWF starts out ahead, then drops off, as shown elsewhere.
*TWF rogues are steady improvement while fighters are spiky--fighters get big boosts at 5, 11, and 20 while sneak attack steadily increases. Overall, they're behind but not by much.
*The assumption that Paladins were designed spend all their slots smiting seems a broken one. Everyone else is consistent near the fighter's numbers (except the archers) and then at level 9 the paladins start taking off like a rocket. Adding an archetype (especially Vengeance) would make some of this worse (but give incentives to spend spell slots on other things than smites). But that's way too much work right now.

If we assume that they only spend half their slots (favoring smiting out of 1st level slots, then 2nd, etc) on smites, the numbers become more reasonable.



Archetype
Level 1
Level 5
Level 9
Level 13
Level 17


Paladin
100
99
103
106
109



So this brings me to a question/thought problem: When designing a class, what should be the benchmark for burst? The GWF fighter? The full-nova paladin? Something else?

JNAProductions
2017-10-05, 08:25 PM
Depends on what the class can do.

The Fighter, for instance, is pretty much pure endurance. Action Surge is its only burst power, excepting Archetypes. If your class has little burst power, it should have strong at-will.

The Wizard, on the other tentacle, is largely burst power. Its at-will suffers a lot, with the best being 4d10 damage a turn, but it can nova like no-one's business.

PhoenixPhyre
2017-10-05, 08:33 PM
Depends on what the class can do.

The Fighter, for instance, is pretty much pure endurance. Action Surge is its only burst power, excepting Archetypes. If your class has little burst power, it should have strong at-will.

The Wizard, on the other tentacle, is largely burst power. Its at-will suffers a lot, with the best being 4d10 damage a turn, but it can nova like no-one's business.

ETA: Turns out that the paladin (after IDS kicks in) actually keeps up with the at-will (no resource) fighter pretty well (~7% lower) without smiting at all. My hunch is that the devs figured that the fighter would be burning an Action Surge about every other fight and the paladin smiting with about half his spell slots. Dunno what they were thinking with the TWF and archery types.

Does it count if the burst is primarily multi-target? I have a subclass that can pull serious burst (basically by spending ki-equivalent to ricochet arrows off of one target into another, but no arrow can hit the same target twice).

The other main subclass is an attempt to create a dedicated TWF subclass that keeps very close to par with the fighter. (The other two are a utility/healer and Captain America, the subclass, but I'm not balancing them against a pure DPR subclass per se). All have a mostly-shared toolkit of non-combat versatility and an anti-caster spin.

It's a spin on the basic fluff of 3.5e's Magic of Incarnum but with totally revamped mechanics.

JNAProductions
2017-10-05, 08:35 PM
I'd say you can have multi-target burst TOTAL DAMAGE be a bit higher (no more than 30%, I say by pulling a number out of my patoot) than single-target burst, but should still not be too much higher.

Morphic tide
2017-10-07, 04:12 AM
...why the hell are you expressing it as a percentage of GWF Fighter? Seriously, why? Flat numbers are more useful for all purposes, here, because it lets us know up front what the overall damage dealt by characters is at various levels, which is far more useful for balance than relative numbers compared to the most basic of the classes that has a reputation for high unconditional damage. I started a thread trying to compile such things. (http://www.giantitp.com/forums/showthread.php?516966) It got called useless because of not including explicitly-optional rules(feats and multiclassing) that complicate calculations quite a lot and transform the question into a serious optimization trial.

Seriously, instead of asking this question and giving ratios, why not show the number crunching for all the classes comparable to what you're trying to do? Then we can discuss what makes for a good balance point and how to reach it.

Edit: As for balance, I think that Eldritch Knight Fighter is a good balance point for Seeker. Both have some damage components locked behind long rests, while also having plenty of at-will damage.

PhoenixPhyre
2017-10-07, 07:20 AM
...why the hell are you expressing it as a percentage of GWF Fighter? Seriously, why? Flat numbers are more useful for all purposes, here, because it lets us know up front what the overall damage dealt by characters is at various levels, which is far more useful for balance than relative numbers compared to the most basic of the classes that has a reputation for high unconditional damage. I started a thread trying to compile such things. (http://www.giantitp.com/forums/showthread.php?516966) It got called useless because of not including explicitly-optional rules(feats and multiclassing) that complicate calculations quite a lot and transform the question into a serious optimization trial.

Seriously, instead of asking this question and giving ratios, why not show the number crunching for all the classes comparable to what you're trying to do? Then we can discuss what makes for a good balance point and how to reach it.

Edit: As for balance, I think that Eldritch Knight Fighter is a good balance point for Seeker. Both have some damage components locked behind long rests, while also having plenty of at-will damage.

I don't have tons of time right now, but I chose the GWF as a reference point because its damage is high but steady. Using the rogue (which has 0 burst) as the balance point doesn't really change the trends much, although the numbers change. That's another column in the spreadsheet below. Ratios are nice because relative error (% changes) are more important and more readable (to me at least) than raw numbers. A 1.0 DPR difference is very different if the average is 5 DPR than if the average is 500 DPR. Ratios account for this and allow easier tracking of trends. Also, large tables full of decimal numbers get hard to read. This is something I learned while getting my degree.

This post isn't even really about the Seeker--it still needs a lot more work before that point. This was just a result of doing the associated numerical work. It's designed to be suggestive, not comprehensive, and point out some interesting trends.

BTW, I think that your idea was unfairly maligned. Balancing around high-op (feats/multiclassing) is an exercise in futility that only leads to power creep. In addition, it's absurdly hard to do.

As far as theory goes, I have a strong hunch (that is borne out more and more as I do more number crunching) that the designers did their balance work on the basic rules archetypes. That's why (as well as sheer ease, since I'm terminally lazy) I used the champion fighter as my reference. If a proposed class is within 10% (rough margin) either way, it's probably fine but further adjustments can be done in play-test.

I'm suspicious of trying to balance exactly based on spreadsheet numbers--my rule of thumb is that about 10% variation is fine, 20% may be fine under some circumstances, and more than 20% needs adjustment.

For your reference, here's a view-only link to the spreadsheet I'm using--It's currently a mess without comments. The lines labeled Seeker are my WIP class's results. https://1drv.ms/x/s!AjKe-YTGxfZWrXGGG6MZGIdOAgIE

In the spoiler is the table redone with raw DPR numbers (no-resource DPR / sustainable burst DPR). I've also flipped it (rows for columns). The paladin numbers assume preferentially smiting out of lower slots and using ~1/2 of slots for smites and the rest for spells.




Level
GWF F
TWF F
LB F
TWF R
LCB R
PLD


1
7.1 / --
9.1 / --
6.0 / --
9.5 / --
7.7 / --
7.1 / --


5
18.5 / 20.8
17.9 / 19.5
16.7 / 18.7
16.8 / 16.8
15.0 / 15.0
17.3 / 20.5


9
20.0 / 22.5
20.2 / 22.0
17.1 / 19.2
23.1 / 23.1
21.4 / 21.4
19.2 / 22.4


13
30.0 / 33.7
26.9 / 29.5
25.6 / 28.8
28.7 / 28.7
27.0 / 27.0
26.4 / 34.4


17
31.1 / 38.8
27.6 / 33.0
26.3 / 32.8
34.3 / 34.3
32.6 / 32.6
26.4 / 39.4

Lalliman
2017-10-07, 10:54 AM
...why the hell are you expressing it as a percentage of GWF Fighter? Seriously, why? Flat numbers are more useful for all purposes, here, because it lets us know up front what the overall damage dealt by characters is at various levels, which is far more useful for balance than relative numbers compared to the most basic of the classes that has a reputation for high unconditional damage.
I strongly disagree. The raw numbers are meaningless without a point of comparison. If you have a table of damage numbers, you're just going to compare them to the other classes' damage anyways. Phoenix has made that easier by taking away the interpretation step and giving us the information that matters. I genuinely don't see why the raw DPR numbers would be more effective for answering this particular question.

As for the actual question, you can't really speak of how much burst potential a class "should" have. I would say to just calculate the burst damage into the total sustained damage, and make sure the class doesn't burst any harder than the burstiest core class (probably paladin). JNAP's 30% rule seems intuitively reasonable for multi-target damage.

Morphic tide
2017-10-07, 11:05 AM
I don't have tons of time right now, but I chose the GWF as a reference point because its damage is high but steady. Using the rogue (which has 0 burst) as the balance point doesn't really change the trends much, although the numbers change. That's another column in the spreadsheet below. Ratios are nice because relative error (% changes) are more important and more readable (to me at least) than raw numbers. A 1.0 DPR difference is very different if the average is 5 DPR than if the average is 500 DPR. Ratios account for this and allow easier tracking of trends. Also, large tables full of decimal numbers get hard to read. This is something I learned while getting my degree.
I understand the point of ratios, but my real frustration is the lack of any finite numbers. A simple 100/X for the Fighter row would be great by me, as it would give the exact DPR for the others' ratio just by running the ratio.


BTW, I think that your idea was unfairly maligned. Balancing around high-op (feats/multiclassing) is an exercise in futility that only leads to power creep. In addition, it's absurdly hard to do.
I know. I also got scolded for useless entries, which is something I want to figure out better. But for that, I need to be clear about the things I'm working with, like average hit rates for CR=character level encounters.


As far as theory goes, I have a strong hunch (that is borne out more and more as I do more number crunching) that the designers did their balance work on the basic rules archetypes. That's why (as well as sheer ease, since I'm terminally lazy) I used the champion fighter as my reference. If a proposed class is within 10% (rough margin) either way, it's probably fine but further adjustments can be done in play-test.
The balance work is on no feats, no multiclassing and probably basic archetypes. Minimal optimization to be had, well defined bounds to work with. Feats and multiclassing are explicitly called optional rules, so they are probably not balanced for intensely.


I'm suspicious of trying to balance exactly based on spreadsheet numbers--my rule of thumb is that about 10% variation is fine, 20% may be fine under some circumstances, and more than 20% needs adjustment.
Oh, that's perfectly fine by me. The main reason I made that thread for getting the DPR done was to get the numbers on classes so that I could balance a subsystem's baseline around existing classes. The subsystem in question being mana pools with per-round recovery in the form of d4s, and the ability to just flatly blow points from the pool for effects. Temp HP, outright healing, damage of various types and generally being able to have level-appropriate damage continuously, having actual spells use up points in such a ratio that only AoE with two or more targets outdoes just flatly spending points on effects.


In the spoiler is the table redone with raw DPR numbers (no-resource DPR / sustainable burst DPR). I've also flipped it (rows for columns). The paladin numbers assume preferentially smiting out of lower slots and using ~1/2 of slots for smites and the rest for spells.




Level
GWF F
TWF F
LB F
TWF R
LCB R
PLD


1
7.1 / --
9.1 / --
6.0 / --
9.5 / --
7.7 / --
7.1 / --


5
18.5 / 20.8
17.9 / 19.5
16.7 / 18.7
16.8 / 16.8
15.0 / 15.0
17.3 / 20.5


9
20.0 / 22.5
20.2 / 22.0
17.1 / 19.2
23.1 / 23.1
21.4 / 21.4
19.2 / 22.4


13
30.0 / 33.7
26.9 / 29.5
25.6 / 28.8
28.7 / 28.7
27.0 / 27.0
26.4 / 34.4


17
31.1 / 38.8
27.6 / 33.0
26.3 / 32.8
34.3 / 34.3
32.6 / 32.6
26.4 / 39.4



Thanks. I really need to work out more coherent categories for the setup I want the classes to be crunched for, because the point is ultimately to give a direct comparison of capabilities so that classes aiming for particular biases can be balanced properly. The utter strictness of 5e's encounters-to-rests ratio is quite helpful, here.