Seeing RED--an odd coincidence

[PhoenixPhyre]

RED here is "Rogue Equivalent Damage", a unit I'm fiddling with. Basically, take the most dead simple rogue you can think of--a shortsword, always has sneak attack, uses their bonus action every turn for something else. Always hits, but never crits[1]. Starts at DEX = +3 and increases it at every opportunity until capped (level 8).

Basically, this is the most braindead damage unit you can use. It's not intended to be particularly representative of anything real, just an abstract scaling unit for use in other ways. No resources to worry about, no fancy math. No assumptions about fight length or number of rounds per day--it's only dependent on modifiers and levels.

But that's not really the point of this point, which is to point out an odd "coincidence" I noticed while doing some fiddling with the monster numbers I've generated. It was spurred by the idea of calculating Rounds To Kill against a single (averaged) monster of a particular CR, using the average HP of monsters in the (old[2]) core books at each of those CRs. But what CRs to pick?

Originally, I said "CR = Level". Because that's easy, right? But rogue damage only goes up every other level, but monster HP increases every level (some). So you got this periodic pattern of starting below trend, then increasing, then resetting.

So I thought...well, if RED increases basically every other level most of the time...why not compare to CR = Level / 2 (rounding down)? At higher levels, this matches the expected median CR of mass encounters using Xanathar's guidance for encounters, so it's not impossible (even if it's very outside the norm for most tables).

When I did that, I got a very nice trend. To one decimal place (because that's all the precision I really have here), the rounds to kill are between 1.9 (level 1) and 3.4 (level 10, CR 5), and basically (with a very few exceptions), alternate between being slightly below trend to slightly above trend, and the fit gets better as levels increase. The average is 2.9 over all levels.

If I were to make grand claims, I'd say that it's consistent with a model where the "3 round" idea for calculating CR comes from assuming
a) the averaged, after all the noise washes out, per-PC damage is roughly 1 RED
b) and the average foe is roughly CR = level / 2.

Do I think that this is anything but a nice coincidence? Not really. But it's consistent with that very simple model. Which was intriguing.

[1] oddly enough, if you just naively include crits (hits and misses, so damage becomes 0.9*normal + 0.05*crit), damage for that simpleminded rogue goes down, not up. The cost of losing 5% of hits is greater than the extra damage from crits. Which makes sense, since crit damage isn't double the total, but only double the dice. If you included a more realistic accuracy number up front, I don't think that would hold. But I'm too lazy to do the math.
[2] Ie MM, VGtM, and MToF[4], not MM and MotM.

[kazaryu]

[1] oddly enough, if you just naively include crits (hits and misses, so damage becomes 0.9*normal + 0.05*crit), damage for that simpleminded rogue goes down, not up. The cost of losing 5% of hits is greater than the extra damage from crits. Which makes sense, since crit damage isn't double the total, but only double the dice. If you included a more realistic accuracy number up front, I don't think that would hold. But I'm too lazy to do the math.
.

no math required. a more realistic accuracy number would already include the miss chance of rolling a 1, so you're correct that it wouldn't hold up. crit chance is just a straight increase in DPR (not a good one, mind, but still only a benefit.

[Chronos]

I'm not sure what the coincidence was, here... You found that the rounds to kill a monster varied with level. Well, of course it varied with level; what would you expect? Is the "coincidence" just that it varied smoothly? That just means that RED as a function of level and average monster HP as a function of level are both smooth functions.

[Tanarii]

I think the point was it didn't significantly vary with level, as long as you assume CR = 1/2*PC level. But it's hard to since there's not accompanying chart of the results. :smallamused:

The basic point here being that it's a VERY common assumption when thinking about the system math to compare to CR = PC level, and this significantly distorts things by even mid Tier 2, but definitely by Tier 3 or Tier 4. Personally I've always used CR = level-3 as a comparison point.

However, the basic flaw in this RED method seems to be "always hits". It's not clear from the post that when compared to CR to determine RTK, this is then adjusted for that CR's typical AC.

[Greywander]

How does this compare to the warlock baseline of EB + AB? That's usually the metric I see getting used to determine where a build is on the DPR curve. Also, I've heard it suggested that a 65% hit rate is roughly what's expected. As your attack bonus goes up, so does monster AC, so it stays pretty close to 65%. At least, that's what I've heard.

[PhoenixPhyre]

I think the point was it didn't significantly vary with level, as long as you assume CR = 1/2*PC level. But it's hard to since there's not accompanying chart of the results. :smallamused:

The basic point here being that it's a VERY common assumption when thinking about the system math to compare to CR = PC level, and this significantly distorts things by even mid Tier 2, but definitely by Tier 3 or Tier 4. Personally I've always used CR = level-3 as a comparison point.

However, the basic flaw in this RED method seems to be "always hits". It's not clear from the post that when compared to CR to determine RTK, this is then adjusted for that CR's typical AC.

As expected, doing the full accuracy calculation introduces substantial noise. However, accuracy is basically random across those level bands, varying from 63% (CR 1/Level 2) to 82% (CR 8/Level 17) but not monotonically at all. (Note, this is done by comparing to the average AC for each of the selected CRs).

The Rounds to Kill metric actually becomes flatter (and slightly larger), with an average value of 3.5 and extrema of (2.5@level 1, 4.0@level6&10), but the confidence in a linear fit goes way down.

And posting pictures here is obnoxious. It's all in the spreadsheet in my signature, the tab labeled RED Investigation. There are a lot of other numbers in that tab as well, because it's mostly scratch space.


How does this compare to the warlock baseline of EB + AB? That's usually the metric I see getting used to determine where a build is on the DPR curve. Also, I've heard it suggested that a 65% hit rate is roughly what's expected. As your attack bonus goes up, so does monster AC, so it stays pretty close to 65%. At least, that's what I've heard.

Note that 1 RED is not a baseline for the DPR curve for any build. It's just an easy calculation to benchmark more difficult ones against. EB + AB is substantially lower in a 100% accuracy, no crits case (which is what I have time for before work, because doing accuracy for multi-attack stuff takes work). Ranging from 0.55 RED at level 1 (no AB) up to 0.85 RED at level 2, but then bouncing around averaging 0.67 RED. The variance will be much smaller than for RED once accuracy is considered due to multi-attack considerations (post level 5 at least). RTK varies from 2.9@level 5 to 5.5@level 10, averaging 4.3

Effectively, 1 RED is a touchpoint (not a metric, but a touchpoint) for resourceless damage. If your resourceless damage is substantially below 1 RED (like a warlock's is), then I expect to see a significant amount of resource-using damage. Or at least the capability of doing so. However, if your resourceless damage is close to (on either side) of 1 RED, I expect to see less resource-dependency for damage. But the value of the metric in the greater case is still in question, as I need to do more crunching for various different cases.

[Jak]

So, I'm not really following on the CR part, but RED is a nice sounding acronym, and the baseline I was using a while back when building a monk. I'm curious about the metric; what are you using it to measure against?
It seems more realistic appropriate than the warlock-ag-hex baseline.

[Tanarii]

I've always found accuracy vs CR-3 AC to be a fairly stable value. But I also don't assume that chars will start with an attack stat of 16 nor that they increase their attack stat immediately at 4 and 8 on top of a 16 starting point.

[KorvinStarmast]

How does this compare to the warlock baseline of EB + AB? *Grinds teeth* That isn't a baseline.

[Greywander]

*Grinds teeth* That isn't a baseline.
??? Not sure what you mean by that. Any time you want to talk about a build being good or bad you need to have something to compare it to. For damage, warlock is as good a comparison as any. EB + AB is good, but not great, so if you're doing less damage than that it means your damage probably isn't great. If you're doing more damage than that, you're probably doing pretty well.

Now, damage isn't everything, so raw DPR doesn't really tell you if a build is effective or not, and I think that contributes to why monks, for example, are chronically underrated. But if you are looking at damage specifically and want to know where you stand, the warlock comparison is functional. It is a baseline, but not the only one, and not necessarily the best one.

Like, we could argue whether the Imperial system or the Metric system is better, but both can still be used to measure things. If people are getting useful information out of the warlock comparison, I'm not sure why that would upset you.

[PhoenixPhyre]

??? Not sure what you mean by that. Any time you want to talk about a build being good or bad you need to have something to compare it to. For damage, warlock is as good a comparison as any. EB + AB is good, but not great, so if you're doing less damage than that it means your damage probably isn't great. If you're doing more damage than that, you're probably doing pretty well.

Now, damage isn't everything, so raw DPR doesn't really tell you if a build is effective or not, and I think that contributes to why monks, for example, are chronically underrated. But if you are looking at damage specifically and want to know where you stand, the warlock comparison is functional. It is a baseline, but not the only one, and not necessarily the best one.

Like, we could argue whether the Imperial system or the Metric system is better, but both can still be used to measure things. If people are getting useful information out of the warlock comparison, I'm not sure why that would upset you.

I think (and he can correct me if I'm wrong) that what KorvinStarmast meant was that it's an entirely arbitrary, unpinned from anything in the text or system baseline. Which is true. It's basically setting the bar wherever you want. Which is fine, but not all that useful for anything. It's akin to saying "the right baseline is 42".

Personally, I prefer not to use baselines (which assume that if you're below that, you're "doing it wrong"). Or at least if I do, I set them very low, to where (as far as I can tell) the system expects it. So a halfling rogue wielding a greatclub with 0 STR is below the baseline and "doing it wrong" (as far as expected DPR is concerned). Because it's obviously anti-optimizing. So the truest baseline for each class is a single-classed, SRD version with the "obvious" choices. Ie following the quick build for each class, bumping your main attack stat at each opportunity until capped, and using proficient armor and weapons appropriately sized for your race. Yes, that's a very low bar. But it's the system's bar. Which is what matters.

RED is not intended (once it's fully fleshed out) to be a baseline. Instead it's just a (mostly arbitrary) metric for resource-free damage. A "scaling factor" if you will. And more for ideating on classes, not on builds. Because there's too much variance in builds. But basically, if a class has resource-free damage that is

a) << 1 RED, it needs substantial and common resources that can deal damage to make up (some of) the difference.
b) >> 1 RED, it's probably too strong. especially if it has damaging resources to employ to spike its damage higher.
c) ~ 1 RED (+- a fairly wide range here, like 30% either way), it needs more inspection and playtest.

My personal issue with the EB + AB "baseline" is if you include hex. Because hex isn't a resource-free option, especially for warlocks. Nor is it a set-it-and-forget-it option, even at higher levels (due to concentration). So setting the baseline as EB + AB + hex is putting the thumb on the scale quite a lot and causes massive distortions.

[Witty Username]

How does this compare to the warlock baseline of EB + AB? That's usually the metric I see getting used to determine where a build is on the DPR curve. Also, I've heard it suggested that a 65% hit rate is roughly what's expected. As your attack bonus goes up, so does monster AC, so it stays pretty close to 65%. At least, that's what I've heard.

Generally speaking RED will be a little lower, but will grow at a different rate generally more evenly across levels.
Warlock damage will spike at roughly at 5th, 11th and 17th with a bump at 2nd level when they take agonizing blast, (and I would recommend hex not being factored in until the warlock has 2 spell slots, which is also 2nd level).
Rogue has a more unique damage growth, at 1d6 every odd level, with no spikes in output for the most part.

Now for an example, lets compare at 1, 2, 3, and 5 a Fighter (let's use a greatsword), Rogue (RED), and Warlock(EB + Hex):
1st
fighter 2d6+3, rogue 2d6+3, warlock 1d10
avg, fighter 10, rogue 10, warlock 5.5
2nd
fighter 2d6+3, rogue 2d6+3, warlock 1d10+3+1d6 (AB and Hex)
avg, fighter 10, rogue 10, warlock 12
3rd
fighter 2d6+3, rogue 3d6+3 (sneak bump), warlock 1d10+3 +1d6
avg, fighter 10, rogue 13.5, warlock 12
5th
fighter 4d6+8 (extra attack), rogue 4d6+4(sneak bump), warlock 2d10+8+2d6 (extra blast)
avg, fighter 22, rogue 18, warlock 26

This cross section of tier 1 shows us that that the rogue, RED, is generally lower than the warlock baseline for the most part, but also spikes above in 3rd level because of that uniform growth of rogue damage.

If you want to use a damage baseline, rogue will generally be easier to beat than the warlock baseline, and will have different stress points on the build. Eyeballing RED, 9th, and 15th levels are likely the hardest to beat for most martial characters.

[PhoenixPhyre]

Generally speaking RED will be a little lower, but will grow at a different rate generally more evenly across levels.
Warlock damage will spike at roughly at 5th, 11th and 17th with a bump at 2nd level when they take agonizing blast, (and I would recommend hex not being factored in until the warlock has 2 spell slots, which is also 2nd level).
Rogue has a more unique damage growth, at 1d6 every odd level, with no spikes in output for the most part.

Now for an example, lets compare at 1, 2, 3, and 5 a Fighter (let's use a greatsword), Rogue (RED), and Warlock(EB + Hex):
1st
fighter 2d6+3, rogue 2d6+3, warlock 1d10
avg, fighter 10, rogue 10, warlock 5.5
2nd
fighter 2d6+3, rogue 2d6+3, warlock 1d10+3+1d6 (AB and Hex)
avg, fighter 10, rogue 10, warlock 12
3rd
fighter 2d6+3, rogue 3d6+3 (sneak bump), warlock 1d10+3 +1d6
avg, fighter 10, rogue 13.5, warlock 12
5th
fighter 4d6+8 (extra attack), rogue 4d6+4(sneak bump), warlock 2d10+8+2d6 (extra blast)
avg, fighter 22, rogue 18, warlock 26

This cross section of tier 1 shows us that that the rogue, RED, is generally lower than the warlock baseline for the most part, but also spikes above in 3rd level because of that uniform growth of rogue damage.

If you want to use a damage baseline, rogue will generally be easier to beat than the warlock baseline, and will have different stress points on the build. Eyeballing RED, 9th, and 15th levels are likely the hardest to beat for most martial characters.

But including hex in that comparison makes it utterly inapposite for comparing to RED or the baseline fighter. Because hex requires resources that aren't guaranteed (or even likely) to be up all day, especially at lower levels. RED is for zero resource damage. Including things with durations introduces (unstated and widely varying) assumptions about number of rounds per fight, number of fights per day, and short rests. Which is extremely out of scope for the RED metric.

If you compare no hex EB + AB, you get something like what I did above. Where it's consistently lower (but the difference is noisy) than 1 RED for true resource-free damage. Which makes sense, warlocks have resources. If you spend them to deal damage, you get much closer (or generally above) 1 RED. Which is normal.

[Greywander]

I think (and he can correct me if I'm wrong) that what KorvinStarmast meant was that it's an entirely arbitrary, unpinned from anything in the text or system baseline. Which is true. It's basically setting the bar wherever you want. Which is fine, but not all that useful for anything. It's akin to saying "the right baseline is 42".

Personally, I prefer not to use baselines (which assume that if you're below that, you're "doing it wrong"). Or at least if I do, I set them very low, to where (as far as I can tell) the system expects it. So a halfling rogue wielding a greatclub with 0 STR is below the baseline and "doing it wrong" (as far as expected DPR is concerned). Because it's obviously anti-optimizing. So the truest baseline for each class is a single-classed, SRD version with the "obvious" choices. Ie following the quick build for each class, bumping your main attack stat at each opportunity until capped, and using proficient armor and weapons appropriately sized for your race. Yes, that's a very low bar. But it's the system's bar. Which is what matters.

RED is not intended (once it's fully fleshed out) to be a baseline. Instead it's just a (mostly arbitrary) metric for resource-free damage. A "scaling factor" if you will.
Right, but I think the warlock baseline fulfills a similar purpose. I don't really see any argument that makes logical sense that criticizes using the warlock baseline but praises using RED. They're both basically the same thing, but based off of a different class.

The warlock baseline is also very similar to the fighter baseline, since both add an extra attack in each tier. The fighter baseline would be calculated without a fighting style or subclass, since those can and will influence DPR in nuanced ways that are hard to account for. 1d8 + STR/DEX mod, or 1d10 + CHA mod, a number of times equal to your tier, is a pretty simple and easy calculation to use.

And yes, if you're trying to build a damage dealer, and you can't beat these baselines, then you are doing it wrong. If you're not trying to build a damage dealer, then who cares? Control and support can offer a lot more value to the party than extra damage would.


My personal issue with the EB + AB "baseline" is if you include hex. Because hex isn't a resource-free option, especially for warlocks. Nor is it a set-it-and-forget-it option, even at higher levels (due to concentration). So setting the baseline as EB + AB + hex is putting the thumb on the scale quite a lot and causes massive distortions.
I don't include Hex, just like I wouldn't include a fighting style or subclass on the fighter baseline. There are any number of ways to boost damage, but all of them are optional. Some fighters may choose a defensive fighting style instead of an offensive one, for example. Some (many) warlocks will prefer to cast a control spell instead of Hex. By figuring those into the baseline, we essentially are telling players they're doing it wrong if they don't use those options.

Anyway, I think this RED concept is pretty cool, it's another tool we can use. I guess the warlock baseline should then properly be called "WED", and the fighter baseline "FED". We could probably also get "BED", "MED", and "PED", and then rangers can cry over RED already being taken. I don't think full casters other than warlocks would work for this, though, and warlocks only work because they lean so hard on EB.

[Witty Username]

I don't include Hex, just like I wouldn't include a fighting style or subclass on the fighter baseline. There are any number of ways to boost damage, but all of them are optional. Some fighters may choose a defensive fighting style instead of an offensive one, for example. Some (many) warlocks will prefer to cast a control spell instead of Hex. By figuring those into the baseline, we essentially are telling players they're doing it wrong if they don't use those options.

Anyway, I think this RED concept is pretty cool, it's another tool we can use. I guess the warlock baseline should then properly be called "WED", and the fighter baseline "FED". We could probably also get "BED", "MED", and "PED", and then rangers can cry over RED already being taken. I don't think full casters other than warlocks would work for this, though, and warlocks only work because they lean so hard on EB.

Ah, my calc included hex as that is what I have observed most people use, in that case I think RED does generally beat it in damage output. I think 5th level is an exception.


Full casters can use cantrips, 1d8 is probably the fairest one but d10 will have no riders. it will be lower than rogue damage at all levels.

[PhoenixPhyre]

Right, but I think the warlock baseline fulfills a similar purpose. I don't really see any argument that makes logical sense that criticizes using the warlock baseline but praises using RED. They're both basically the same thing, but based off of a different class.

The warlock baseline is also very similar to the fighter baseline, since both add an extra attack in each tier. The fighter baseline would be calculated without a fighting style or subclass, since those can and will influence DPR in nuanced ways that are hard to account for. 1d8 + STR/DEX mod, or 1d10 + CHA mod, a number of times equal to your tier, is a pretty simple and easy calculation to use.

And yes, if you're trying to build a damage dealer, and you can't beat these baselines, then you are doing it wrong. If you're not trying to build a damage dealer, then who cares? Control and support can offer a lot more value to the party than extra damage would.


I don't include Hex, just like I wouldn't include a fighting style or subclass on the fighter baseline. There are any number of ways to boost damage, but all of them are optional. Some fighters may choose a defensive fighting style instead of an offensive one, for example. Some (many) warlocks will prefer to cast a control spell instead of Hex. By figuring those into the baseline, we essentially are telling players they're doing it wrong if they don't use those options.

Anyway, I think this RED concept is pretty cool, it's another tool we can use. I guess the warlock baseline should then properly be called "WED", and the fighter baseline "FED". We could probably also get "BED", "MED", and "PED", and then rangers can cry over RED already being taken. I don't think full casters other than warlocks would work for this, though, and warlocks only work because they lean so hard on EB.

Ah. Without hex then yeah, it's fairly similar. WED would be smaller and much more "lumpy" than RED--instead of nice smooth boosts every other level you get a big boost every 5 or so. Which is why I like RED most. Especially since it's dead simple to include accuracy in RED, while doing it for multi-attackers is...less so. Also, rogues basically have no "class choices" for damage once you fix a sub-class to something like Thief. Fighters you have to worry about weapon choice, GWF vs SnB, sub-classes (most of whom have resources), the class itself having resources...Basically, RED is as true-to-life as you get--you can actually perform at or very near RED as a rogue without neglecting a huge chunk of your class. Such as, say, a ranged short-bow rogue who takes his bonus action to do the whole advantage thing (either via hiding or the Tasha's ACF). The other metrics require ignoring a chunk of your class.


Ah, my calc included hex as that is what I have observed most people use, in that case I think RED does generally beat it in damage output. I think 5th level is an exception.

Full casters can use cantrips, 1d8 is probably the fairest one but d10 will have no riders. it will be lower than rogue damage at all levels.

WED without hex is lower at all levels. The high is at levels 5 and 6, where it hits 0.83 RED. And once you include accuracy, a pure-cantrip caster is always going to be significantly below 1 RED--warlocks will be the best of the pack. And without AB it's not even a contest--pure EB sits between 0.38 RED and 0.55 RED. As would, say, a wizard casting firebolt as their only damage. In this approximation, more dice is more dice whether more attacks or not, while if you include accuracy, the value of a crit goes down (each attack is smaller) but the probability of hitting at least once goes way up, as does the chance of scoring a crit during your turn. Without number crunching, I'd guess that more attacks (even without modifiers) is, overall, better. But it's not as good as it seems.

In fact, the big draw of PAM isn't getting a bonus action 1d4 attack. It's getting a bonus action 1d4 + modifier attack. Take that away and it goes way down in value.

[Greywander]

Full casters can use cantrips, 1d8 is probably the fairest one but d10 will have no riders. it will be lower than rogue damage at all levels.
Full casters don't use cantrips all that often, though. There's almost always a better leveled spell you could be using instead. This is why I don't think full casters work very well for this kind of thing; it's just very difficult to determine an accurate baseline when the class is built primarily around using limited burst resources instead of sustainable damage. I think a better metric for burst damage would be things like the maximum damage they can dump out in a single turn, or over a few rounds. The whole point of burst damage is to drop a target as fast as possible so you don't have to deal with them anymore.


Ah. Without hex then yeah, it's fairly similar. WED would be smaller and much more "lumpy" than RED--instead of nice smooth boosts every other level you get a big boost every 5 or so. Which is why I like RED most.
Yeah, if you want a level by level analysis then RED works well, since it's constantly scaling. WED/FED is better for comparing whole tiers. Some classes may get a damage boost at 7, 8, 10, or 14, but a lot of classes only get one major damage boost per tier, such as paladins getting Extra Attack then Improved Divine Smite. RED is more granular, but you don't always need that level of detail.


Especially since it's dead simple to include accuracy in RED, while doing it for multi-attackers is...less so.
Not really? Just take the average damage of each hit, add them together, then multiply by the hit rate, e.g. 0.65 for a 65% hit rate. Or you can just use LudicSavant's damage calculator (https://forums.giantitp.com/showthread.php?582779-Comprehensive-DPR-Calculator-(v2-0)).

For example, average tier 4 warlock damage if all beams hit is 4d10 + 20 = 22 + 20 = 42. With a hit rate of 65%, the expected DPR is 42 * 0.65 = 27.3. The damage calculator can also figure in crits and such, giving a more accurate DPR of 28.4. Tier 4 rogue DPR is 30.2, assuming a d6 weapon like a shortsword.


Also, rogues basically have no "class choices" for damage once you fix a sub-class to something like Thief. Fighters you have to worry about weapon choice, GWF vs SnB, sub-classes (most of whom have resources), the class itself having resources...Basically, RED is as true-to-life as you get--you can actually perform at or very near RED as a rogue without neglecting a huge chunk of your class. Such as, say, a ranged short-bow rogue who takes his bonus action to do the whole advantage thing (either via hiding or the Tasha's ACF). The other metrics require ignoring a chunk of your class.
This is also true of warlocks, as outside of a few things, such as Hex, Hexblade's Curse, or Genie's Wrath, there aren't a lot of ways to boost EB damage beyond what AB gets you.

But yeah, FED would be a misleading baseline, since fighters have a lot more options for increasing damage, including just getting more ASIs that can be spent on feats. Same for PED, as ignoring smites will severely lowball paladin DPR, but there's not really a good way to guess how often the paladin will smite (pretty much the same issue as with full casters). BED can be more predictable, if we assume the barb is always raging and using Reckless Attack, which is pretty baseline for barbarians.


In fact, the big draw of PAM isn't getting a bonus action 1d4 attack. It's getting a bonus action 1d4 + modifier attack. Take that away and it goes way down in value.
Right, at that point it would just be TWF with extra steps. PAM and CE are really strong in tier 1 where they nearly double your damage output, but drop off a bit later on. They're still better than not having them, of course, but to get the most out of them you need something that utilizes the extra attack. This might be as simple as a rogue getting a second chance to apply Sneak Attack, or something like a paladin adding Improved Divine Smite damage to the BA attack.

[Tanarii]

Ah.
WED without hex is lower at all levels. The high is at levels 5 and 6, where it hits 0.83 RED. And once you include accuracy, a pure-cantrip caster is always going to be significantly below 1 RED--warlocks will be the best of the pack. And without AB it's not even a contest--pure EB sits between 0.38 RED and 0.55 RED. As would, say, a wizard casting firebolt as their only damage. In this approximation, more dice is more dice whether more attacks or not, while if you include accuracy, the value of a crit goes down (each attack is smaller) but the probability of hitting at least once goes way up, as does the chance of scoring a crit during your turn. Without number crunching, I'd guess that more attacks (even without modifiers) is, overall, better. But it's not as good as it seems.
More attacks vs less attacks is exactly the same average damage for a given accuracy, as long as the total damage dice (including on crits) is identical. The only difference is variance, more attacks means a wider spread of damage, more chance of some damage, and less chance of a lot of damage.

Similarly lots of smaller damage dice (as Rogues get) means less variance on hit, even if the average is the same.

Single attack Rogue in particular have an interesting situation where they might have the same average damage as another character with many smaller damaging attacks, but they have a large chance of 0 damage and a large chance of another tightly invariant damage if they hit, whereas the multi attacker could have a wide variation of damage across the total spectrum of possibilities, and a single attack large-but-fewer damage dice character would have a large chance of 0 and a large chance of a higher variance damage if they hit.

Same applies to a non-Agonizing EB vs Firebolt at higher levels. Lots of folks are trained to think EB is pure better, but some folks don't want reliably average damage from multiple EB attacks. They'd rather have (for example) a 40% chance of 0 damage and a 60% chance of 3d10.

[Psyren]

Personally, I prefer not to use baselines (which assume that if you're below that, you're "doing it wrong").

That's not the point of the baseline. The point is to have a common damage number that's relatively easy to hit that we can then use to weigh other things.

If your class is meeting or below the baseline, that doesn't mean you're playing "wrong;" it just means that hopefully your class does other things since pure damage isn't their strength. Lots of great builds don't care about exceeding or the baseline, or even meeting it in some cases.

[PhoenixPhyre]

That's not the point of the baseline. The point is to have a common damage number that's relatively easy to hit that we can then use to weigh other things.

If your class is meeting or below the baseline, that doesn't mean you're playing "wrong;" it just means that hopefully your class does other things since pure damage isn't their strength. Lots of great builds don't care about exceeding or the baseline, or even meeting it in some cases.

That's not what I've ever heard a baseline to mean in anything. I've always heard that if there's a baseline for X, your performance in X should be at least X if you care about it at all. That is, it's the system's expectations. Sure, a pure lore bard isn't likely to care much about a damage baseline. But if your main thing is damage, you should be at or above the system's baseline. And no, the hex + EB + AB is not some "relatively easy to hit" number, precisely because it's a skewed look at things (assuming that you'll always have concentration, etc).

The problem with D&D is that the optimization culture has made it so the "baseline" is drastically inflated compared to the system's baseline. Like in another thread people are saying that a sustained 100 DPR after accuracy is the expectation for any damage dealer at level 20. When that's more than 2 RED. Something that a thief rogue cannot ever approach. Not even close. Not even with a +3 weapon (that adds...3 damage per turn) or even a +3, +2d6 weapon. Advantage doesn't help, because 1 RED already assumes always hitting.

Also, you clipped out the next sentence, where I backed off of "no baseline" to "use the system baseline" (aka "don't anti-optimize, follow the quick build and take what seems good"). That's all the system expects. Anything more than that is gravy. Conveniently, that's basically 1 RED (+- about 10% in many cases). All the basic rules classes? Pretty darn close to 1 RED (previous calculations put most of them running between 0.9 and 1.5 RED) averaged over a full day. Some are spikier, some, like the rogue, are more constant. You only get the really really spiked numbers if you make ludicrous assumptions (one fight per day, a ton of tiny creatures conveniently in fireball formation, mass simulacra, several pre-buffing rounds, etc). And even most of the PHB classes/subclasses are fairly near 1 RED. As I showed, it also ends up that if you have a party of 4 doing, on average, about 4 RED/round you kill 4 CR = Level / 2 monsters in 3 turns. Ish. Which is...right in line with all the guidance from the DMG.

One of the whole aims of the RED formulation is to put this intuition (which I've had for a long time based on experience) on a firmer methodological foundation. It's not a baseline or even a competitive metric (ie something to compare any given build against), but a system-design exploration. If, as I suspect, that's the closest thing to a "fundamental unit of damage per turn" that the game has, it lets me examine the effects of
a) homebrew changes
b) optimization claims (to examine possible houserules, etc)
c) monster design
d) etc.

It's not about optimization bragging rights. That's a topic I have less than zero interest in[1].


and one that I think is utterly corrosive to a cooperative game. The "how high can I push the numbers" game is fine...for an MMO or a competitive game. One where the challenges are fixed. Or for pure idle theorycrafting in one's head. But making it seem like it's the system's expectations and anything less is wrong (and yes, that is exactly how these forums come across a large majority of the time, where if you're not pushing all the numbers you're suspect, and if monsters aren't tuned to the biggest numbers possible they're "weak" and "unworthy" and "badly designed," CF Vecna). I dislike that attitude tremendously.

[Psyren]

That's not what I've ever heard a baseline to mean in anything. I've always heard that if there's a baseline for X, your performance in X should be at least X if you care about it at all. That is, it's the system's expectations.


It's not a system expectation.

I do "care about" my damage as say, a control wizard, but only when I have nothing better to do with my actions. It's not my primary job. If you zoom out and look at my total damage across multiple fights I'm probably going to fall below a Hex+AB+EB baseline - but it's quite possible regardless that I save the group's lives more than once.

If however I'm trying to make damage be my primary schtick, having a basic benchmark for that is useful.Not for the system to measure basic competence, but a number the community can use to compare one build to another.

[Greywander]

I don't think anyone's actually disagreeing with you with regards to RED. I'm the one who brought up the warlock baseline as something comparable to RED, which I always intended to exclude Hex since I think there are better spells a warlock can be concentrating on. You've tried to make a distinction between the two, and I don't think they're different at all. I can't speak for what other people have said regarding baseline damage, and I haven't seen anyone in this thread arguing that anyone below baseline is playing wrong. The only assertion I or others have made to that effect is that if you're specifically speccing for damage, then you should be exceeding RED/WED. Otherwise, it kind of begs the question of why you don't just play a rogue or warlock, if you want to deal damage and those do more damage than you.

But there are other ways to contribute to combat than raw damage, and most builds do have other things besides damage that they can contribute to combat. For example, a thief rogue can grab a healer's kit and patch people up as a BA, or deploy caltrops/ball bearings/oil. They can do that, and still match RED. Treantmonk's god wizard will rarely do any damage at all, unless it's to throw a Fireball to wipe out a group of enemies. The god wizard is focused primarily on control and support, making their party more effective and acting as a force multiplier to the party. That's way more useful than doing a bit of damage. But if all you're contributing is damage, it should be well above RED/WED. I just don't think it's wise to build a character who can't do anything other than deal damage. D&D is a team game, and flexibility is key to being capable of tackling a broad variety of situations.

Heck, one of the team optimization concepts I came up with was one focused almost exclusively on defense (https://forums.giantitp.com/showthread.php?637921-The-Wall-Making-an-undefeatable-party), rather than offense. And I think that team build does a good job of showing how important support abilities are. Damage is important, as without it you'll never defeat the enemy, but anyone can deal at least some damage. The actual damage numbers aren't what's important, it's how many of you are still standing when the dust settles, and how many resources you had to spend.

As a tool for homebrew design, I think RED is great, and I'll likely be using it myself in the future.

[Rukelnikov]

The simple reason why Ago Blast is a very good baseline, is that it only commits 2 levels. So any build trying to do damage that doesn't meet AB's damage would be automatically improved with a 2 lvl dip.

Rogue's Sneak Attack is better for fun stuff like calculating expected number of attacks to kill a creature of a given CR at a given level since it scales more smoothly.

[Tanarii]

Warlock with agonizing isn't a baseline. Warlocks without are.

Baselines are minimums to compare against. It's part of the definition.

[Rukelnikov]

Warlock with agonizing isn't a baseline. Warlocks without are.

Baselines are minimums to compare against. It's part of the definition.

Minimum would be unarmed damage 8 str (or any value between 1 and 11, its the same).

Minimum doesn't help in any way to determine how good damage of a given build is.

Damage you can get with a 2 lvl dip is a pretty useful tool, for the reason I explained above.

[Tanarii]

Minimum would be unarmed damage 8 str (or any value between 1 and 11, its the same).
That's not correct for any class,

However it's correct that even EB isn't the minimum for "warlock". You could choose d8 for chill touch instead. It is the baseline for "EB warlock" though.

Meanwhile, RED *is* a baseline. It's an actual damage minimum.

But adding agonizing to EB is no different from adding a flaming short sword to a rogue. It's not even the baseline for "EB warlock"

[Rukelnikov]

That's not correct for any class,

However it's correct that even EB isn't the minimum for a warlock. You could choose d8 for chill touch instead.

Meanwhile, RED *is* a baseline. It's an actual damage minimum.

But adding agonizing to EB is no different from adding a flaming short sword to a rogue.

Minimum would be unarmed, it would deal less damage, by definition the minimum element of a set is no higher than any other, and an unarmed warlock with 1 to 11 str would deal less damage.

Even RED isn't even minimum for a rogue, cause they could use a dagger, or again, unarmed.

[Psyren]

If we're operating off different definitions of the term "baseline" then this isn't going to go anywhere.

[PhoenixPhyre]

The meaning I've always used/understood distinguishes between a benchmark and a baseline.

Benchmarks are neutral. They're just test metrics against which to rank performance. There's no intrinsic meaning to a score of X on a benchmark, it's just a number. A unit for measuring and comparing.

Baselines are opinionated. It's the base line, against which every conforming implementation should be measured (if its relevant to their operation[1]), and where falling below the baseline means that performance improvements are needed to remain in conformance.

By those definitions, RED is a benchmark. A way to standardize DPR numbers at a particular level, including standardizing assumptions. Just like you could measure temperature in eV[2], but doing so is rather odd for 99% of cases, saying "I can do X DPR" is a bit...vague. Under what assumptions and conditions? Using a benchmark (of which RED is only one, and limited to one specific case, that of resource-free damage) that standardizes the assumptions makes things more comparable. But remains value-neutral. A class scoring 1.32 RED at level 5 is just a class that scores 1.32 RED at level 5.

The baseline for the game (and there can only really be one meaningful baseline IMO, that set by the system's definition of "conforming implementation") is something else. It's not the absolute minimum damage you can do, because having a non-conforming implementation (ie build) is possible. It's the value for which a party that ends up meaningfully below the baseline for DPR will struggle to fight "appropriate" foes, where "appropriate" is defined by the system and its guidelines.

Benchmarks can be entirely internal, without reference to things like monsters. Baselines, because they come from the system itself, will generally involve comparisons to monsters.

My hypothesis (and it's a fairly weak one at this point) is that the 5e's true baseline is roughly 1 RED. A party of four wizards who do nothing but spam firebolt and ray of frost is "non-conforming" and will struggle with most "appropriate" encounters. A basic rules party that doesn't anti-optimize and uses their kit (the fighter and rogue will generally be slightly over 1 RED after including resources and magic items, as will the wizard, while the cleric may or may not be depending on details) will not struggle on average.

[1] A heat pump used only in heating mode doesn't care about the manufacturer's baseline for cooling performance.
[2] or any number of other weird and strange units--my favorite is one that sets the speed of light = the mass of an electron = the charge of an electron = hbar = 1 (unitless). Really convenient for certain atomic physics calculations, really WTF for most other things.

[Nefariis]

Not trying to throw in a wrench or anything - but I feel like Rogue's off-hand attack or Arcane Tricksters' Booming blade cantrip should also be added to the Rogue score, both are resource free. Should/would they be considered in the RED calculation?

[PhoenixPhyre]

Not trying to throw in a wrench or anything - but I feel like Rogue's off-hand attack or Arcane Tricksters' Booming blade cantrip should also be added to the Rogue score, both are resource free. Should/would they be considered in the RED calculation?

The benchmark assumes a Thief rogue (because build choices that can add discretionary damage, such as "does the enemy move" for the BB trigger are against the spirit of a simple benchmark). Adding in an offhand attack would marginally change the values, but again, discretion. The assumption here is really a ranged rogue (shortbow) who always uses their bonus action for something else (usually hiding or the whole increased accuracy thing so that they can guarantee getting sneak attack every turn and always hitting).

So you could add it in, but my preference is to keep it as simple as possible. You could benchmark a TWF or AT/BB rogue against the RED rogue, and I'd expect marginally increased values. But it doesn't change the metric as presented (thief rogue with one shortsword or a shortbow, always hitting, never critting[1], always using bonus action for something else that doesn't deal damage).

[1] There's an alternate "accuracy-adjusted RED" metric which tries to account for accuracy, including crits. But that's skewed by having to assume specific targets (to get an AC value) or by introducing a separate "against targets of AC X" assumption. And benchmarks, IMO, should be independent of things like monsters (which vary tremendously).

[Rukelnikov]

I second not including BB or GFB for the purposes for which RED was devised.

And I agree its a benchmark, and an interesting one, because of its smoothness. And it may even be one of the nombers they had in their heads while balancing the other classes (this I don't feel strongly about, but it definitely sounds plausible)

I disagree though, that a baseline for damage needs a comparison to monsters*, a given builds damage only needs to be compared to other builds damage.

And again, the reason Ago Blast is a good baseline, is that it can be achieved with just 2 levels, it doesn't require a full build. If something as simple as that is outdoing a different build whichs main purpose was damage, then that other build isn't very good at achieving its objective.

* There are caveats, builds that use wild shape, polymorph or monster statblocks to calculate their own damage, make other builds need to compare against monsters, but not as test dummies, which seemed to me to be the implication.

[PhoenixPhyre]

I disagree though, that a baseline for damage needs a comparison to monsters*, a given builds damage only needs to be compared to other builds damage.


In my formulation of what a baseline means, the baselines are provided by the manufacturers (the devs, in this case). And they're always going to be in system terms--not "you should be doing more damage than a XYZ under conditions ABC", but "you should be able to handle what we say are expected encounters". Which necessarily reflects external conditions (ie what the system considers expected encounters). Otherwise it's not a stable baseline and is utterly arbitrary. Which is fine for a benchmark, but not for a baseline (which is there to say "you must be this tall to ride").

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

On a side note--is there a widely-accepted figure for the DPR contribution of the GWF style? As in "on average, with a greatsword, it adds X damage per hit"?

[Rukelnikov]

In my formulation of what a baseline means, the baselines are provided by the manufacturers (the devs, in this case). And they're always going to be in system terms--not "you should be doing more damage than a XYZ under conditions ABC", but "you should be able to handle what we say are expected encounters". Which necessarily reflects external conditions (ie what the system considers expected encounters). Otherwise it's not a stable baseline and is utterly arbitrary. Which is fine for a benchmark, but not for a baseline (which is there to say "you must be this tall to ride").

IMX designers don't design for people that care about things like DPR, thus optimization is not in their equations, thus, whatever baseline they used is pointless. It wouldn't be IF we cared about monsters, but as I said, we don't because we are comparing builds damage

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


On a side note--is there a widely-accepted figure for the DPR contribution of the GWF style? As in "on average, with a greatsword, it adds X damage per hit"?

Well you can calc the average damage of a roll with reroll involved it should be:

Average value for a D sided die, where you reroll up to R number would be:

Avg(D, R) = (R*A + (D-R)*(D-R+1)/2 + R*D - R^2) / D

D = Die sides
R = number up to which you reroll
A = Average value for a D sided fair die = (D*D+1)/2

I'm sure that can be turned into a neater formula, but I really don't feel like doin it rite now, sry. (Note that if you take R = 0 it will give the standard value, goddammit. im already doing crap... gimme a minute)

I think the avg increase for a D sided die with reroll up to number R is:

AvgDmgGain (D, R) = (R*A - R*(R+1)/2)/D

[Greywander]

Baselines are minimums to compare against. It's part of the definition.
Here's the definition from Dictionary.com (https://www.dictionary.com/browse/baseline)


Baseball. the area between bases within which a base runner must keep when running from one base to another.
Tennis. the line at each end of a tennis court, parallel to the net, that marks the in-bounds limit of play.
(in perspective drawing) a horizontal line in the immediate foreground formed by the intersection of the ground plane and the picture plane.
a basic standard or level; guideline: to establish a baseline for future studies.
a specific value or values that can serve as a comparison or control.

Those last two definitions seem to fit how we've been using it, though I'll agree there's some ambiguity. The idea that a baseline is meant to serve as the lowest permissible value does make a certain amount of sense, but I don't think that's how the term always gets used, e.g. in this very thread.


If we're operating off different definitions of the term "baseline" then this isn't going to go anywhere.
Agreed, I feel like several people in this thread are getting stuck on the word itself and ignoring how it's actually being used, even though it's been explained several times what we mean by "baseline".


The meaning I've always used/understood distinguishes between a benchmark and a baseline.
These definitions do make some sense. It's one of the reasons I didn't include Hex for the warlock baseline. The baseline is not the absolute minimum, but it's a lower, unoptimized value. RED and WED are easily achievable by most classes, though some classes will struggle to reach them without optimization because they're not designed to be damage dealers.

Benchmarks, on the other hand, are generally going to be quite a bit higher. The baseline is unoptimized and easy to surpass, while the benchmark is a feat to reach and exceed. Reaching a benchmark requires optimization, as the whole point is to check against another optimized build (which serves as the benchmark) to see if you can do better.

I agree with what Rukelnikov said: Basically, if you're trying to deal damage but can't achieve WED, then a 2 level dip into warlock will at least get you up to WED. Pretty much this is just a more specific version of, "If you're trying to do X, but you can't even do X as well as a generic, unoptimized Y, why don't you just play a Y instead?"

As for a baseline being defined by the system itself, I think there's a lot of nuance there and it might be hard for everyone to agree on what the real baseline is. If you want an objective baseline, then perhaps the closest you can get is to look at what the DMG considers an appropriately balanced encounter at each level and calculate how much DPR is required from a standard-sized party to defeat that encounter in the expected time frame (which might be three rounds, though I'm not sure if the DMG explicitly states this anywhere). This might produce some strange results, though.

I agree with not including BB/GFB in the RED calculation, as that's more of a benchmark than a baseline. Likewise for including Hex on the warlock; that's a benchmark, not a baseline. I do think Agonizing Blast should be included in the warlock baseline, as without it EB spam is little different from any other cantrip spam. Cantrip spam is well below baseline because those classes are expected to be casting leveled spells most of the time, which the warlock can't due to limited pact magic slots.

[KorvinStarmast]

Greywander, I will now suggest that the next time you drag a Treantmonk assumption into a thread, realize that you may not get away with it.

[PhoenixPhyre]

Moving away from the benchmark vs baseline conversation (which I don't really care about other than to keep up my reputation as a language pedant), some numbers for the champion fighter.

Additional assumptions beyond RED:
* Effect of Enhanced Critical is calculated as the extra damage beyond "regular" crits (which aren't included on anyone's side of the ledger): (extra critical chance c)*(damage per attack)*(A + 2(A-1)*c + 3(A-2)*c^2 + 4(A-3)*c^3) where A is the number of attacks and the extra terms are assumed to be MAX(X, 0).
* For Action Surge calculations (ie with consuming resources), assuming 18 rounds (6 fights at 3 rounds each) with a variable number of short rests.

All numbers in units of RED, presented as AVG, [MIN, MAX] (Note Updated as below comments)


Weapon choice
Resource Free
0 short rests
1 short rest
2 short rests
3 short rests
notes


SnB (d8 weapon) + dueling FS
0.70, [0.54, 0.97]
0.75 [0.60, 1.02]
0.7 [0.66, 1.08]
0.85 [0.72, 1.13]
0.89 [0.77, 1.18]
High at level 1, low bounces around and is fairly constant but levels 16-17 tend to be low


GWF style, greatsword, assuming a flat 0.8/hit contribution from GWF
0.97 [0.87, 1.17]
01.04 [0.91, 1.23]
1.06 [0.96, 1.30]
1.12 [1.0, 1.36]
1.19 [1.05, 1.1.43]
High at level 1, low generally levels 8-9



Are these numbers exact? No, I'd have to consider all the crits. But it does give me more confidence in the metric--the two "easy to calculate" basic rules damage dealers are pretty darn close to each other when in damage configuration. And the "tank" configuration is significantly lower (as expected).

Spreadsheet as google doc https://docs.google.com/spreadsheets/d/1IvlhRezlvPEIHnKLTHHilg1p8XJr2tEs3O1kYhb7e_o/edit?usp=sharing

Can I guarantee my numbers are correct? Heck no. But I think they're reasonable.

[Rukelnikov]

Moving away from the benchmark vs baseline conversation (which I don't really care about other than to keep up my reputation as a language pedant), some numbers for the champion fighter.

Additional assumptions beyond RED:
* Effect of Enhanced Critical is calculated as the extra damage beyond "regular" crits (which aren't included on anyone's side of the ledger): (extra critical chance c)*(damage per attack)*(A + 2(A-1)*c + 3(A-2)*c^2 + 4(A-3)*c^3) where A is the number of attacks and the extra terms are assumed to be MAX(X, 0).
* For Action Surge calculations (ie with consuming resources), assuming 18 rounds (6 fights at 3 rounds each) with a variable number of short rests.

All numbers in units of RED, presented as AVG, [MIN, MAX]


Weapon choice
Resource Free
0 short rests
1 short rest
2 short rests
3 short rests
notes


SnB (d8 weapon) + dueling FS
0.50, [0.30, 0.97]
0.53 [0.34, 1.02]
0.57 [0.37, 1.08]
0.60 [0.4, 1.13]
0.63 [0.44, 1.18]
almost monotonic decrease from level 1


GWF style, greatsword, assuming a flat 0.8/hit contribution from GWF
0.93 [0.77, 1.13]
0.99 [0.86, 1.19]
1.06 [0.91, 1.26]
1.12 [0.96, 1.32]
1.19 [1.00, 1.38]
much more varied with peak moving around but most frequently being levels 5 and 6



Are these numbers exact? No, I'd have to consider all the crits. But it does give me more confidence in the metric--the two "easy to calculate" basic rules damage dealers are pretty darn close to each other when in damage configuration. And the "tank" configuration is significantly lower (as expected).

Spreadsheet as google doc https://docs.google.com/spreadsheets/d/1IvlhRezlvPEIHnKLTHHilg1p8XJr2tEs3O1kYhb7e_o/edit?usp=sharing

Can I guarantee my numbers are correct? Heck no. But I think they're reasonable.

I didn't check all the numbers, I'm about to go to sleep, but the 0.8 contribution can't be right, the contribution of rerolling up to 2 in a d6 is:

(2*3.5 - 2*3/2) / 6 = 4/6 = 0.666...

in a greatsword it'd be 8/6 = 1.333...

[PhoenixPhyre]

I didn't check all the numbers, I'm about to go to sleep, but the 0.8 contribution can't be right, the contribution of rerolling up to 2 in a d6 is:

(2*3.5 - 2*3/2) / 6 = 4/6 = 0.666...

in a greatsword it'd be 8/6 = 1.333...

:shrug: I just took what I found on a quick google search. Probably for a different weapon or something. Thanks for the correction.

Edit: that changes that particular row's averages to

No resources: 0.97
0 SR: 1.04
1: 1.11
2: 1.18
3: 1.25

[PhoenixPhyre]

After another correction from @Rukelnikov, the SnB numbers change to (averages):

base: 0.7
0 SR: 0.75
1: 0.8
2: 0.85
3: 0.89

and the change is no longer montonic but bounces around more. I'd forgotten to multiply by the number of attacks (oops).

[Witty Username]

Why isn't sneak attack considered discretionary damage?

[PhoenixPhyre]

Why isn't sneak attack considered discretionary damage?

Because that makes it really hard to calculate :smallwink:

And from the other numbers, it seems that "you always get sneak attack if you hit" is a fairly safe assumption. Sure, it's artificial, but it's way less artificial than being dependent on things like number of rounds/long rest, number of short rests, number of monsters, discretionary things like "holding smites for a crit" or "picking which slots to use" or widely varying things like "how long can I keep hex up."

Basically, I can't get away from all assumptions, but that seems like a reasonably safe assumption. And the rest of the data for the basic rules classes bears it out as a reasonable assumption--the champion fighter falls right in line with it. And the cantrip-only numbers for casters (presented below) also bear it out, especially for warlocks (EB + AB).

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

Evocation wizard (firebolt only): Average 0.58 RED, (0.42 - 0.73 )(low at level 9 to high at level 10, where they get to add modifier to a spell attack).

Warlock (EB without AB) or non-evocation wizard (firebolt): Average 0.50 RED (0.38 - 0.55).

Cleric, Sacred Flame + Potent Spellcasting: Average 0.51 RED (0.31 - 0.63)

Warlock (AB + EB): Average 0.9 (0.55 - 1.07) (with the low being there at level 1 when AB isn't online)

So the "traditional" full casters do about 50% of an RED with cantrips alone, with evocation wizards doing a tiny bit better at high levels (10+) and potent spellcasting making up for lower (basic rules) cantrip damage, barely, after level 8.

AB + EB spam gives you basically 1 RED, with pulses higher or lower (higher when you just got new beams and lower right before you get new beams).

Edit: Also did the TWF rogue (shortsword/shortsword) case: Average 1.16 (so basically the same as the 2SR greatsword champion at 1.18) (1.35@level 1 - 1.08@level 20), with a monotonic decrease because that extra 3.5 damage diminishes as a % of the total. As expected. This reinforces the measure's reasonability--the "damage focused" builds are all coming in right about the same ball-park.

[Witty Username]

Hm, did a quick calculation of bladesinger
Scimitar and firebolt.
And got a high estimate of about .85 RED (high estimate as I only calculated 6, 11, and 17 which is the most favorable levels for bladesinger and there is likely to be an accuracy disparity that these won't account for as dex is likely to be increasing for rogue but is much less likely for bladesinger, and bladesinger has two stats important for basic damage).
With a damage of 1.0 RED at 6th.

[PhoenixPhyre]

Hm, did a quick calculation of bladesinger
Scimitar and firebolt.
And got a high estimate of about .85 RED (high estimate as I only calculated 6, 11, and 17 which is the most favorable levels for bladesinger and there is likely to be an accuracy disparity that these won't account for as dex is likely to be increasing for rogue but is much less likely for bladesinger, and bladesinger has two stats important for basic damage).
With a damage of 1.0 RED at 6th.

That tracks. It's quite a bit higher than the evocation wizard spamming cantrips, and accounting for the stat disparity makes the difference "in the noise" as they say. 6th is also a weak point for rogues, since it's right between bumps.

I expect that the actual bulk of the damage comes from full spells. Which is a right royal pain to calculate with lots of baked-in assumptions. In large part because you have to account for the loss of a cantrip each round you cast a full spell--fighters with Action Surge don't, since it's strictly additive. So I'm not going to do that right now, because lazy.

[PhoenixPhyre]

Here's an odd one--what about an evocation wizard whose combat actions are either
a) using a spell slot for magic missile (starting with highest level slots)
b) casting firebolt
(still under the "always hits, never crits" assumption, which does devalue magic missile slightly)?

Assumptions: 18 rounds of combat with at least one short rest (used to do Arcane Recovery, always prioritizing more slots over bigger slots, at least if that will give more castings of MM).

Note: For levels 9+, they will only cast MM, having at least 18 spell slots (including AR).

The outcome strongly depends on the interpretation of magic missile--do we go with the One Damage Roll interpretation (Empowered Evocation applies to every bolt) or the Multiple Damage Roll interpretation (Empowered Evocation only applies to one bolt per casting).

ODR: Damage is nearly constant at ~0.62 RED for levels 1-9. At level 10 (when Empowered Evocation comes online), damage immediately spikes to 1.47 RED and averages 1.36 RED for the next 10 levels, more than doubling. The lowest damage point in those top ten levels is 1.28 RED; the highest of the previous was 0.67 RED. Global average 1.03 RED, but basically bimodal.

MDR: Everything's the same below level 10. At level 10, it spikes to 0.8 RED and then declines back to previous levels fairly quickly after that. Global average...0.67 RED. So basically flat.

------thoughts on this weird case
1. Wow, that was annoying to calculate. And changing the number of rounds would alter things fairly drastically and require almost total recalculation (because there are more resources than rounds after level 9). And different assumptions about how to spend AR will force complete recalculation. Ugh.
2. I don't think this is a really realistic scenario (who does nothing but cast magic missile except a Nuclear Wizard?), but it does show a significant effect fairly clearly.

------Other thoughts/questions:
If you were a wizard, what spell of each level would be your "go to" for single-target blasting? As in, if you had to spend a slot of that level to deal damage to a single target but could pick any wizard spell, what would you choose? Including upcasting lower level ones if that makes sense.

[Rukelnikov]

------thoughts on this weird case
1. Wow, that was annoying to calculate. And changing the number of rounds would alter things fairly drastically and require almost total recalculation (because there are more resources than rounds after level 9). And different assumptions about how to spend AR will force complete recalculation. Ugh.
2. I don't think this is a really realistic scenario (who does nothing but cast magic missile except a Nuclear Wizard?), but it does show a significant effect fairly clearly.

I'll try n write a script for that, it shouldn't be that difficult (I hope)


------Other thoughts/questions:
If you were a wizard, what spell of each level would be your "go to" for single-target blasting? As in, if you had to spend a slot of that level to deal damage to a single target but could pick any wizard spell, what would you choose? Including upcasting lower level ones if that makes sense.

As a blanket statement? Without knowing what the oppossition could be? Then MM, its likely the singlemost reliable single target damage in the game.

[PhoenixPhyre]

I'll try n write a script for that, it shouldn't be that difficult (I hope)


Yeah. Scripting would definitely be the way to go for the more dynamic ones. You can automate the "turns" fairly well with a loop. I'll probably switch to a proper script setup for any other calculations for that reason

[ff7hero]

Off the top of my head, I'd say Wizard damage spells would be Magic Missile at Level 1, Scorching Ray Levels 2-4 and Animate Objects at Level 5+. If AO complicates things too much, Scorching Ray is probably still a decent baseline. Above Level 5 I'm not really sure about.

[Witty Username]

------Other thoughts/questions:
If you were a wizard, what spell of each level would be your "go to" for single-target blasting? As in, if you had to spend a slot of that level to deal damage to a single target but could pick any wizard spell, what would you choose? Including upcasting lower level ones if that makes sense.

Probably magic missile, scorching ray, then file not found.
I think blight and disintegrate are the only single target damage spells that are beyond 2nd level I have any memory of and it don't particularly like either of them.

Magic missile is good IMO though mostly because it cannot miss, it is often insufficient but rarely a waste.

[Cheesegear]

But that's not really the point of this point, which is to point out an odd "coincidence" I noticed while doing some fiddling with the monster numbers I've generated. It was spurred by the idea of calculating Rounds To Kill against a single (averaged) monster of a particular CR, using the average HP of monsters in the (old[2]) core books at each of those CRs. But what CRs to pick?

I had a similar idea when trying to decide how necessary character optimisation even is. If you take the DMG custom monster rules as given, and that everything officially published falls inside the guidelines of the DMG, you end up with this, assuming that the Guidelines in Xanathar's are also correct:

https://i.imgur.com/DGjtKTo.png

If your DPR is somewhere between the blue and red lines, your character should do just fine. You'll notice that monster HP spikes when player damage does, interestingly enough. Levels 4-5, 11-12, and 16+. I don't know if this was intentional. But it's really ****ing clever if it was.

Just for clarity; The blue line does not represent 'bad' DPR. Below the blue line is bad DPR.

If your DPR is above the red line, whether or not you're power-gaming (such as you roll a crit on your down and double/triple your damage), this frees up some of your party to fail their attacks, and/or not deal damage on their turn (e.g; Healing). Which is a fun lesson to teach. Having one character dealing a ****-ton of damage isn't necessarily overshadowing other members of the party, one party member doing ****-tons of damage, actually frees up the other members of the party to not do damage at all. But of course whether this matters or not depends entirely on your players' mindset, and creative encounter design - in that there wont always be things to do in combat outside of dealing damage, and maybe the players don't want to do those things, even if there are.

Now you can plug your REDs into that graph to see if RED is even good or not.

[PhoenixPhyre]

I had a similar idea when trying to decide how necessary character optimisation even is. If you take the DMG custom monster rules as given, and that everything officially published falls inside the guidelines of the DMG, you end up with this, assuming that the Guidelines in Xanathar's are also correct:

https://i.imgur.com/DGjtKTo.png

If your DPR is somewhere between the blue and red lines, your character should do just fine. You'll notice that monster HP spikes when player damage does, interestingly enough. Levels 4-5, 11-12, and 16+. I don't know if this was intentional. But it's really ****ing clever if it was.

Just for clarity; The blue line does not represent 'bad' DPR. Below the blue line is bad DPR.

If your DPR is above the red line, whether or not you're power-gaming (such as you roll a crit on your down and double/triple your damage), this frees up some of your party to fail their attacks, and/or not deal damage on their turn (e.g; Healing). Which is a fun lesson to teach. Having one character dealing a ****-ton of damage isn't necessarily overshadowing other members of the party, one party member doing ****-tons of damage, actually frees up the other members of the party to not do damage at all. But of course whether this matters or not depends entirely on your players' mindset, and creative encounter design - in that there wont always be things to do in combat outside of dealing damage, and maybe the players don't want to do those things, even if there are.

Now you can plug your REDs into that graph to see if RED is even good or not.

A couple data points: at level 20, 1 RED is 43 DPR. Which sits comfortably above the blue line. I'd have to not be on a phone to see if it's above the red. But note that's pre accuracy--including accuracy probably (without doing the math) puts it near the blue.

The other tier breakpoints are 10 dpr, 18 dpr, 30 dpr, and 40 dpr.

And yes that spike at CR 5 , etc is very obvious at least in the data for published monsters. If it wasn't intentional, it's one heck of a coincidence.

If you have the raw data for what CRs you looked at for each party level, I could do an accuracy adjusted RED and overlay it on the graph for comparison.

[Cheesegear]

If you have the raw data for what CRs you looked at for each party level, I could do an accuracy adjusted RED and overlay it on the graph for comparison.

https://i.imgur.com/IUH4xCg.png

Limitations in that it doesn't factor in AC of the hostile.

But it is good to know that if your Level 3 Barbarian crits with their Greataxe dealing 20+ damage using Reckless Attacks, the party's Cleric can spend their turn healing, because the Barbarian did enough damage to stay ahead of the curve for the both of them. The Barbarian is going to need the heals, 'cause he's about to get slammed by the Veteran with three attacks who has advantage on every attack.

[Tanarii]

A couple data points: at level 20, 1 RED is 43 DPR. Which sits comfortably above the blue line. I'd have to not be on a phone to see if it's above the red. But note that's pre accuracy--including accuracy probably (without doing the math) puts it near the blue.
Accuracy should drop RED to about 60% of its former value.

If RED doesn't include accuracy, it's not DPR, because DPR includes accuracy.

[Cheesegear]

Accuracy should drop RED to about 60% of its former value.

Partial disagree. Bounded accuracy in the first place, means that modifiers to hit between characters are fairly negligible. The only time you ever need to talk about accuracy is when you're dealing with Power Attack or Sharpshooter.

Accuracy for everyone else, not using those two specific Feats (or similar Feats I just don't know off the top of my head), is more or less the same. Which is why I'm okay with ignoring accuracy.
If the Fighter with 18 Str would hit, the Wizard with 18 Int would probably hit, too. So why introduce accuracy just to confound things, when it actually cancels itself out?

[Rukelnikov]

Tanarii is right.

RED doesn't take accuracy into account, whether 60% is a reasonable number or not is debatable, but I don't think it strays very far from expected, what Tanarii points out is that usually PCs would lose about 40% DPR when you factor accuracy in. The Rogue in particular doesn't fall for this if they make more than one attack a turn, since the bulk of their damage is their Sneak Attack.

A Champion missing one attack at lvl 5 loses half its damage, a Rogue missing one attack at lvl 5 only loses ~35% if missing first hit, and only ~13% if missing off hand. (didn't factor in crits tbh, so its probably a bit off)

If we factored in a 60% hit chance and crits, on top of the Non acc damage:

Champion goes:
4d6+8 = 22 =ACC=> 13.9 ==> ~63% NonACC damage

While rogue:
5d6+4 = 21.5 =ACC=> 16.4175 ==> ~76% NonACC damage

So acc doesn't hit everyone evenly, but it still does hit everyone, and for anoticeable ammount if we use values usually deemed reasonable or in this case 60%

So if RED is around the damage benchmark you posted, the base thief used to calc RED would be below the blue line (at least at level 20 which was the one singled out) , which takes us to the next point...

A- How do you determine what is low and High DPR? Nevermind that, why do you decide that a single point of DPR turns you from low to high?

If you are comparing it against published monsters, as I said before, that's meaningless from a character building perspective, monsters and PCs belong in different sets, you shoudn't compare them as if they were from the same. Compare a given build's damage against other possible build's damage when trying to determine wether DPR is low or high. I'll assume by low you mean "enough" and by high you mean "more than enough", that's different and might be ok if following the guidelines.

However, the assumption that if Char A does X damage above "expected" or "enough" then char B "can" spend their turn not doing damage is wrong on a couple levels.

First, putting aside all roleplaying aspects, considering it as if it were just a wargame, If A dealt a lot of damage, then B dealing more damage on top of that won't be a waste, unless all the monsters are dead, as long the enemy is standing damage will be useful (for most cases were damage can be useful at all). If by some reason 3 characters are prevented from taking their action, Char D dealing 4x damage, would still likely be worse than Char D freeing everyone else from whatever is preventing them from acting. Its not a game where you have damage quotas that must be met on a round by round basis.

Second, it IS a roleplaying game, you don't need permission to do X or Y.



2WF Thief

5% 82
--5% 7
--55% 3.5
--40% 0

55% 43.5
--5% 7
--55% 3.5
--40% 0

40% 0
--5% 77
--55% 38.5
--40% 0

12d6+5 = 47 =ACC=> 39.4 ==> ~84% NonACC damage

Greatsword Champion

15% 19
45% 12
40% 0

8.25 *4 = 33

8d6+20 = 48 =ACC=> 33 ==> ~69% NonACC damage

Greatsword Fighter, non-champion

5% 19
55% 12
40% 0

7.55 *4 = 30.2

8d6+20 = 48 =ACC=> 30.2 ==> ~63% NonACC damage

[Cheesegear]

RED doesn't take accuracy into account, whether 60% is a reasonable number or not is debatable

It is debatable, and that's why we have to ignore it for baseline discussions.


If you are comparing it against published monsters, as I said before, that's meaningless from a character building perspective, monsters and PCs belong in different sets, you shoudn't compare them as if they were from the same.

Okay, my DPR calculations (not PP's RED), comes from a place of 'Do you need to optimise your character? To the degree that the internet kind of tells you to.' The answer, according to my very generous calculations, is 'No.'


I'll assume by low you mean "enough" and by high you mean "more than enough", that's different and might be ok if following the guidelines.

That is correct.

If you're a Level 5 Human Fighter with Dual-Wielding Hand Crossbows and outputting 40+ damage per turn...That's simply not necessary. And if you feel like it is necessary, and any Fighter you play at Level 5 that isn't churning out 40 damage per turn is a waste of time...I've got news for you; The game just isn't that hard. It's just not. I've got a graph that says doing 120 damage in three rounds, on your own, kind of isn't necessary, and if you feel like there's something else you'd rather take for your build...You can! Because sooner or later I'm going to put you in a situation where Dual Wielding Hand Crossbows wont solve the situation, and I hope your character is well-rounded enough to handle that. Are you totally sure you want to be locked into this build...'Cause I'm telling you that you don't have to be, and more, it's not necessary to be.


However, the assumption that if Char A does X damage above "expected" or "enough" then char B can spend their turn not doing damage is wrong on a couple levels.

No it isn't.
If Character B wants to deal damage, they should feel well within their rights to do so. There is very clearly benefits to doing 'excessive' damage - the hostile dies faster.

However, if Character B has a choice between dealing damage, and doing...Anything else...Character B knows that Character A has done 'enough' damage for both of them, the fight is still on track, and it wont necessarily hurt, if Character B takes an action that doesn't deal damage.

If Character A just crits and smashes for massive damage - 'enough' for two characters' worth of DPR.
Character B makes an attack...And misses. Dealing zero damage. Do you panic? No...Everything is fine.

That's where I'm coming from.

How much damage is enough damage? How do you know if your build is at least playable? That's what my graph, and my data is for. My data is for comparing your, individual build, against the monsters you're likely to see in the game if your DM is playing fair. My data is not for comparing your, individual build, to another player's... Or even to any other build you've ever seen. Because 'Who does more damage in this cooperative game?' is well, asinine.

D&D is not that hard. Here's how many HPs hostiles have at each level, and here's an amount of damage you should aim for. Having less damage isn't recommended, but, that being said, having more damage than the red line isn't needed. You don't need a 16 in your primary stat for your character to work. A 16 in your main stat is nice to have, but you should never, ever feel like you need a 16 or your character is unplayable...Here, look at this graph I can prove it...Just play whatever you want...It's okay...

And that's what I'm trying to combat with my data;
Can you fully optimise your character to the point where you outshine other members of the party? You can, if you want.
Do you need to fully optimise your character, possibly to the point where your character is incapable of doing anything outside the specific task of doing massive damage? ...No. D&D isn't like that, not really.

[Boci]

Do you need to fully optimise your character, possibly to the point where your character is incapable of doing anything outside the specific task of doing massive damage? ...No. D&D isn't like that, not really.

Surely you can't know that without knowing the DM. I've played with at least a few "optimize or die" DMs (or rather "optimize or watch your party member die, since you won't be the primary target"). Now what I don't know is whether or not that's because they liked this playstyle, or if it was simple because they tended to have players who would optimize and by the time I joined they had already set the difficulty level high enough that I wouldn't be able to contribute meaningfully without at least a fair bit of optimization.

[Rukelnikov]

How much damage is enough damage? How do you know if your build is at least playable? That's what my graph, and my data is for. My data is for comparing your, individual build, against the monsters you're likely to see in the game if your DM is playing fair. My data is not for comparing your, individual build, to another player's... Or even to any other build you've ever seen. Because 'Who does more damage in this cooperative game?' is well, asinine.

This is where we disagree, I dont think the MM is representative, I almost never play with the monsters as they are presented, as player or DM, unless in an encounter that is of a FAR higher threat than the party is suppossed to be facing.


D&D is not that hard. Here's how many HPs hostiles have at each level, and here's an amount of damage you should aim for. Having less damage isn't recommended, but, that being said, having more damage than the red line isn't needed. You don't need a 16 in your primary stat for your character to work. A 16 in your main stat is nice to have, but you should never, ever feel like you need a 16 or your character is unplayable...Here, look at this graph I can prove it...Just play whatever you want...It's okay...

And that's what I'm trying to combat with my data;
Can you fully optimise your character to the point where you outshine other members of the party? You can, if you want.
Do you need to fully optimise your character, possibly to the point where your character is incapable of doing anything outside the specific task of doing massive damage? ...No. D&D isn't like that, not really.

No matter if you outshine everyone else by a lightyear or if you pick your nose every encounter, the DM will most likely end up balancing premade encounters against the party's capabilities, and when the party are the initiators they should be capable of gauging their own strenght, if they aren't they'll soon learn caution, and thus becaome better at gauging their own capabilities.

[Cheesegear]

Surely you can't know that without knowing the DM.

Maybe you can't.

How effective is your character? How do you know that?

Is it possible to determine your character's effectiveness before you've ever even rolled a dice? How? What metrics should you use? What metrics could you use?

Is another player at your table power-gaming and outshining you, or are you just bad-at-games, and that's why they're outshining you? How do you determine if someone is power-gaming? What even is a 'bad build'?

How sub-optimal can your build be, before it's actual trash? Even against a fair DM.

Why does everyone say Monk sucks? Even against a fair DM.

How come multiple people can say the same thing(s) across the entire internet? Surely they don't play at the same table(s)? Surely they don't have the same DM(s)? But there are conclusions that you, I and they can totally make - outside of a DM - that are simply borne out in maths and reading the rules.

[Rukelnikov]

Maybe you can't.

How effective is your character? How do you know that?

Is it possible to determine your character's effectiveness before you've ever even rolled a dice? How? What metrics should you use? What metrics could you use?

Is another player at your table power-gaming and outshining you, or are you just bad-at-games, and that's why they're outshining you? How do you determine if someone is power-gaming? What even is a 'bad build'?

How sub-optimal can your build be, before it's actual trash? Even against a fair DM.

Why does everyone say Monk sucks? Even against a fair DM.

How come multiple people can say the same thing(s) across the entire internet? Surely they don't play at the same table(s)? Surely they don't have the same DM(s)? But there are conclusions that you, I and they can totally make - outside of a DM - that are simply borne out in maths and reading the rules.

Because saying Monks sucks is DM agnostic*, Monks are a class which can be used for PC builds, in order to gauge how capable they are youmeasure them against other PC builds.

If the weakest PC possible one shotted everything ever published, it would still be the weakest PC posible.

* I don't necessarily agree with the statement though.

[Boci]

How come multiple people can say the same thing(s) across the entire internet?

But that's the point. Multiple, not everyon.e Not everyone agrees monk sucks.

I've heard monk are not the best, but in my game the monk is holding their own with the ranger and warlock and they only just realized stunning fist isn't a bonus action last session. They haven't been using it until now because of that.

[Tanarii]

Leaving out accuracy is fine for inter-character comparisons unless you're talking about something that involves advantage / disadvantage or -5 to hit. It's just not DPR. And if you're trying to compare RED to DPR needed to kill something in a certain number of rounds, you have to modify RED for accuracy.

And it'll be a fairly significant drop if the Rogue doesn't have advantage, less significant if they do. At a quick glance, that means the "expected" DPR at high levels is out of bounds for most single target damage sources unless something like GWM/SS 'power attacks' and advantage are used.

Also rounds to kill CR ~= APL is mostly useful for boss battles. I thought CR ~= 1/2 APL (or APL-3) is what's relevant to discussions about system math? Or is this kind of a reverse engineering, trying to figure out how WotC determined appropriate hit points for boss CRs?

[Chronos]

Just for a baseline of what "baseline" means, see Belkar's alignment chart (https://www.giantitp.com/comics/oots0489.html): Clearly, the Giant just uses it as "something to compare to", not anything like a minimum reasonable value.

[Tanarii]

Just for a baseline of what "baseline" means, see Belkar's alignment chart (https://www.giantitp.com/comics/oots0489.html): Clearly, the Giant just uses it as "something to compare to", not anything like a minimum reasonable value.
The Giant uses lots of terms wrong. /shrug

This really only matters in that some folks, in particular treantmonk, are historically tryng to compare a resource using build to non-resource using ones, using a term that is usually used to mean 'minimum',, and specifically using the term with that meaning in mind to claim other builds are sub-par.

From the perspective of "how does RED compare to EB+AG" it doesn't particularly matter.

From the perspective of calling EB+AG = WED, it very much matters. Because that includes an optimization and value judgement assumption.

[JNAProductions]

The Giant uses lots of terms wrong. /shrug

This really only matters in that some folks, in particular treantmonk, are historically tryng to compare a resource using build to non-resource using ones, using a term that is usually used to mean 'minimum',, and specifically using the term with that meaning in mind to claim other builds are sub-par.

From the perspective of "how does RED compare to EB+AG" it doesn't particularly matter.

From the perspective of calling EB+AG = WED, it very much matters. Because that includes an optimization and value judgement assumption.

Eldritch Blast with Agonizing Blast doesn't use any rest-based resources.
Nor is it dependent on outside factors-you just have it, if you take it.

It's a fine baseline to use. It's not an absolute minimum (that'd be, as mentioned upstream, an 11- Strength Unarmed Strike) but as a "If you want damage to be a thing you're good at, you should be here or better," it's fine.

[PhoenixPhyre]

Eldritch Blast with Agonizing Blast doesn't use any rest-based resources.
Nor is it dependent on outside factors-you just have it, if you take it.

It's a fine baseline to use. It's not an absolute minimum (that'd be, as mentioned upstream, an 11- Strength Unarmed Strike) but as a "If you want damage to be a thing you're good at, you should be here or better," it's fine.

I'll note that it's also below 1 RED, at least before accuracy is included.

I use RED because it's simple and scales smoothly. But that's relatively arbitrary and for convenience. And because I'm single, so I can't use WED. :small-smirk:

[Tanarii]

Eldritch Blast with Agonizing Blast doesn't use any rest-based resources.

I did't write "rest-based" resources.
Invocations are a resource.

[Boci]

I did't write "rest-based" resources.
Invocations are a resource.

Sure, but they're also class features. If rogue had a class feature to boost weapon damage they would likely take it. Rogues get 0 eldritch invocations or anything of the equivalent, so deny warlock an invocation to boost damage is basically denying them part of their class features.

You can argue is bad game design that AB out performs more interesting or dynamic options, but for pure comparison, if a rogue gets all their class features, so does a warlock, and blaster warlocks are going to be taking AB. Taking the eldritch blast cantrip is also a cost, warlocks only get 2 cantrips until level 4.

[Greywander]

From the perspective of calling EB+AG = WED, it very much matters. Because that includes an optimization and value judgement assumption.

I did't write "rest-based" resources.
Invocations are a resource.
The unfortunate truth is that this is just how the warlock is designed. AB isn't a choice, it's a tax. Not taking AB is like playing a polearm rogue: both should be fun and viable builds, but the mechanics simply don't allow it to be so. I think rogues should be able to Sneak Attack with any weapon. I also think that warlocks should have viable alternatives to EB + AB. Imagine, for example, if you could take Vicious Mockery as a cantrip, and an invocation allowed you to target a second or third creature. That would be dope (and on point for GOO/Fathomless 'locks). But no, we don't get anything like that.

So yes, AB takes build resources. But so does simply playing a rogue. Once you've spent levels on rogue, Sneak Attack is free thereafter. Likewise, once you've spent an invocation on AB, the extra damage is free thereafter. You can choose not to take AB (and oh how I wish that was viable), but I would worry that you might not have fun playing that character. Likewise, you might have a really cool idea for a polearm rogue, but I just think it won't be any fun to play due to the mechanics of the game. What I'm saying here is that I think the system is at fault for not supporting these choices, not that the players are at fault for wanting to make those choices.

(I know melee warlocks are also a thing, but the thing is that it actually takes a lot of investment to simply get up to where EB + AB is, though IIRC melee warlocks have the potential to get even higher damage with further investment, whereas EB + AB doesn't really have anywhere to go for more damage. Hex is an option, but concentration is much more precious than a single invocation.)

Anyway, all that to say, I think AB can still be considered a "default" invocation choice. If we were to calculate PED, then we would obviously include Improved Divine Smite, even though the paladin could choose to wield a longbow (and might have good reasons to at certain times, e.g. against flying enemies). Polearm rogues and longbow paladins are going to be rare, and definitely deviate from how those classes are "intended" to be played by the developers. I think the same is true for warlocks who don't take AB. And that's a shame.

Oh well, nothing a bit of homebrew can't fix, right?

[PhoenixPhyre]

For the purposes of my calculations, I'm going to calculate both with and without AB.

It's why I like RED as a touch point -- no choices means it's a stable reference point. It describes a perfectly valid playstyle. That of a thief rogue with a shortbow. Whereas choosing WED as there touchpoint gets into these arguments.

[PhoenixPhyre]

So I've spent most of the day implementing a script-based, web-page version of this idea.

Changes from the OP:
* I've included crits (hits and misses) by default.
* I've included a selector for accuracy with the following options (AC from DMG tables, dex save bonuses from my monster tables, averaged for CR, all flat rolls):
** Default accuracy. 90% hit, 5% crit (unless you have an improved critical feature), 100% fail chance for dex saves.
** "Boss" accuracy (CR = level +3).
** "Half-level" accuracy (CR = level /2 round down).
** "Equal-level" accuracy (CR = level)

I have a preset for the following cases:
* Baseline (guaranteed to end up as 1 RED): Shortbow thief rogue, 100% sneak attack. I haven't properly implemented the Thief's Reflexes feature.
* TWF thief rogue (2 short swords)
* SnB Champion Fighter, Dueling style, with resource options [none | 1SR/3 rounds | 1SR/4 rounds | 1SR/6 rounds | 1 SR/9 rounds]
* GS Champion Fighter, GWF style, with the same resource options
* EB Warlock (no subclass ATM), with options (AB, no AB, EB + 100% hex uptime)
* Cleric, Blessed Strikes, sacred flame with resource options [none | 100% sacred weapon (level 2) uptime]
* Cleric, Potent Spellcasting, sacred flame with the same resource options.

It presents things in graphical as well as tabular format, with comparison (ie you can select multiple presets and graph them together and compare rows in the table).

Still a lot to do, both in presets and allowing advantage/disadvantage. But it's stable, I think, enough for other people to see.

The domain is my personal domain, and it's entirely client-side javascript. Nothing leaves the client except the requests for the js files. It's ugly as sin, to be sure, but I'm a dev, not a designer.
https://admiralbenbo.org/red-calculator/calculator.html

[stoutstien]

So I've spent most of the day implementing a script-based, web-page version of this idea.

Changes from the OP:
* I've included crits (hits and misses) by default.
* I've included a selector for accuracy with the following options (AC from DMG tables, dex save bonuses from my monster tables, averaged for CR, all flat rolls):
** Default accuracy. 90% hit, 5% crit (unless you have an improved critical feature), 100% fail chance for dex saves.
** "Boss" accuracy (CR = level +3).
** "Half-level" accuracy (CR = level /2 round down).
** "Equal-level" accuracy (CR = level)

I have a preset for the following cases:
* Baseline (guaranteed to end up as 1 RED): Shortbow thief rogue, 100% sneak attack. I haven't properly implemented the Thief's Reflexes feature.
* TWF thief rogue (2 short swords)
* SnB Champion Fighter, Dueling style, with resource options [none | 1SR/3 rounds | 1SR/4 rounds | 1SR/6 rounds | 1 SR/9 rounds]
* GS Champion Fighter, GWF style, with the same resource options
* EB Warlock (no subclass ATM), with options (AB, no AB, EB + 100% hex uptime)
* Cleric, Blessed Strikes, sacred flame with resource options [none | 100% sacred weapon (level 2) uptime]
* Cleric, Potent Spellcasting, sacred flame with the same resource options.

It presents things in graphical as well as tabular format, with comparison (ie you can select multiple presets and graph them together and compare rows in the table).

Still a lot to do, both in presets and allowing advantage/disadvantage. But it's stable, I think, enough for other people to see.

The domain is my personal domain, and it's entirely client-side javascript. Nothing leaves the client except the requests for the js files. It's ugly as sin, to be sure, but I'm a dev, not a designer.
https://admiralbenbo.org/red-calculator/calculator.html

Fascinating. Thanks for tossing that together (I'm doing good to get Excel to cooperate most of the time).
It's close to the CDT (certain defeat threshold) rating I'm using ATM. It's a rough estimate on how many actions an NPC is going to get that will impede the party's goal. In the end I'm hoping to have a sliding scale that helps with determining the adjusted challenge of a scenario based on when in the recovery cycle it occurs as much as it's individual rating.

It highlights just how much swing SR based recovery resources have on output which is one of the big things tables struggle to strike a reasonable balance with.

[Chronos]

RED involves choices just as much as WED does. Yes, there are opportunity costs to picking Agonizing Blast, in that picking it means that you're not picking something else. But there are also opportunity costs to picking rogue, or any other specific class, in that it means you're not picking some other class. Mechanically, there would be no difference from the current rules if warlock were split into two different classes, one which got Agonizing Blast and one which got their choice of some other invocation instead.

[Frogreaver]

Off the top of my head, I'd say Wizard damage spells would be Magic Missile at Level 1, Scorching Ray Levels 2-4 and Animate Objects at Level 5+. If AO complicates things too much, Scorching Ray is probably still a decent baseline. Above Level 5 I'm not really sure about.

I'd have went with chromatic orb and help action from familiar.

[Psyren]

Eldritch Blast with Agonizing Blast doesn't use any rest-based resources.
Nor is it dependent on outside factors-you just have it, if you take it.

It's a fine baseline to use. It's not an absolute minimum (that'd be, as mentioned upstream, an 11- Strength Unarmed Strike) but as a "If you want damage to be a thing you're good at, you should be here or better," it's fine.


Sure, but they're also class features. If rogue had a class feature to boost weapon damage they would likely take it. Rogues get 0 eldritch invocations or anything of the equivalent, so deny warlock an invocation to boost damage is basically denying them part of their class features.

You can argue is bad game design that AB out performs more interesting or dynamic options, but for pure comparison, if a rogue gets all their class features, so does a warlock, and blaster warlocks are going to be taking AB. Taking the eldritch blast cantrip is also a cost, warlocks only get 2 cantrips until level 4.

I'm with these two, I see WED as just as useful a baseline if not more, but I'm fine checking builds against RED as well.

[PhoenixPhyre]

I'm with these two, I see WED as just as useful a baseline if not more, but I'm fine checking builds against RED as well.

WED is lumpier (scaling way less frequently) and non-trivially lower. My big issue with it is that it's even more artificial than RED--you're leaving resources that can be used for damage on the table without rationalization. Unlike a shortbow rogue who can't use his bonus action to raise damage much (in the accuracy doesn't matter approximation). It also seems that other basic builds are more closely benchmarked to RED, as they hover really close to 1 RED with dedicated strikers just above, while they'd be substantially above 1 WED.

But sure, you could use WED as a reference point.

[Catullus64]

Sorry to bring this up relatively late - especially since it seems like this metric has caught on in a number of other threads - but why is our assumed Rogue using a shortsword instead of a rapier? If two-weapon fighting is not being considered, there's no real reason not to.

It's an extremely minor point, and, I think, not too disruptive, since it's only a single static increase to the baseline, and doesn't impact scaling. Is it too late to add that +1 average damage to the base for unit calculation?

[PhoenixPhyre]

Sorry to bring this up relatively late - especially since it seems like this metric has caught on in a number of other threads - but why is our assumed Rogue using a shortsword instead of a rapier? If two-weapon fighting is not being considered, there's no real reason not to.

It's an extremely minor point, and, I think, not too disruptive, since it's only a single static increase to the baseline, and doesn't impact scaling. Is it too late to add that +1 average damage to the base for unit calculation?

Because, really, I'd shifted to standardizing on a shortbow rogue instead (leaving the TWF rogue as a comparison). Why shortbow? Because
a) it's easy to calculate (damage is Xd6 + mod)
b) it's more consistent in the no-accuracy limit (it explains why there's no bonus action damage)
c) because I was too lazy to go back and change everything to have a d8 instead :smalltongue:

But now that I've moved to a script-based (instead of spreadsheet-based) calculation method, I can do that change more easily.

Comparing the TWF (dual shortswords) rogue vs the rapier-RED gives an average of 1.11 RED compared to 1.18 RED in the shortbow-RED case. I'll fiddle around and see what I think.

[Catullus64]

Because, really, I'd shifted to standardizing on a shortbow rogue instead (leaving the TWF rogue as a comparison). Why shortbow? Because
a) it's easy to calculate (damage is Xd6 + mod)
b) it's more consistent in the no-accuracy limit (it explains why there's no bonus action damage)
c) because I was too lazy to go back and change everything to have a d8 instead :smalltongue:

But now that I've moved to a script-based (instead of spreadsheet-based) calculation method, I can do that change more easily.

Comparing the TWF (dual shortswords) rogue vs the rapier-RED gives an average of 1.11 RED compared to 1.18 RED in the shortbow-RED case. I'll fiddle around and see what I think.

Points A and C are both valid enough reasons to stick with shortsword as the RED base; it's your show, after all. For B, I would point out that Light Crossbows are also a thing, so the d8 base can still be in place without changing the BA assumption.

[Nefariis]

@PhoenixPhyre

I have a question about your Monk SR/Round calculations.

Looking at this chart -

https://freeimage.host/i/capture.wwoKqQ
https://freeimage.host/i/wwoKqQ

Why would the SR/3 be higher than the SR/4 at level 20?

If I'm understanding the concept, SR/3 implies that the Monk gets a Short Rest after three rounds of combat - but a level 20 monk has 20 Ki points so it's not like he would ever run out.

wouldn't the RED be the same starting at level 8 (2 attacks x 4 rounds)?

[PhoenixPhyre]

@PhoenixPhyre

I have a question about your Monk SR/Round calculations.

Looking at this chart -

https://freeimage.host/i/capture.wwoKqQ
https://freeimage.host/i/wwoKqQ

Why would the SR/3 be higher than the SR/4 at level 20?

If I'm understanding the concept, SR/3 implies that the Monk gets a Short Rest after three rounds of combat - but a level 20 monk has 20 Ki points so it's not like he would ever run out.

wouldn't the RED be the same starting at level 8 (2 attacks x 4 rounds)?

Thanks for catching that. Definitely an error. I'll upload a new calculator version real here soon now, but the short version is that monk overall damage (in the no-accuracy limit, assuming quarterstaff) ends up between 1.35 RED (1/3 rounds) and 1.29 RED (1/9 rounds), with the differences going away after level = # of rounds/SR. Same basic overall pattern (high around level 5-ish, then slowly declining).

Edit: new version pushed. Now also can display raw dpr or % hit (not including crits) for selected cases. No guarantee that the "Constant advantage" case is actually correct--advantage/disadvantage is still experimental.

[Nefariis]

Thanks for catching that. Definitely an error. I'll upload a new calculator version real here soon now, but the short version is that monk overall damage (in the no-accuracy limit, assuming quarterstaff) ends up between 1.35 RED (1/3 rounds) and 1.29 RED (1/9 rounds), with the differences going away after level = # of rounds/SR. Same basic overall pattern (high around level 5-ish, then slowly declining).

Edit: new version pushed. Now also can display raw dpr or % hit (not including crits) for selected cases. No guarantee that the "Constant advantage" case is actually correct--advantage/disadvantage is still experimental.

I think there is a bug in your new version - https://ibb.co/xHKLgMV



calculator.html:84 Uncaught TypeError: Util.getPresetOptions is not a function
at HTMLDocument.<anonymous> (calculator.html:84:18)
(anonymous) @ calculator.html:84

calculator.html:1 Uncaught (in promise) {message: 'A listener indicated an asynchronous response by r…age channel closed before a response was received'}message: "A listener indicated an asynchronous response by returning true, but the message channel closed before a response was received"[[Prototype]]: Object
Promise.then (async)
(anonymous) @ content_script_bundle.js:100
handleNewFeatures @ content_script_bundle.js:93
handleUpdatedNodes @ content_script_bundle.js:93
(anonymous) @ content_script_bundle.js:93
childList (async)
eval @ jquery.js?1157:6253
domManip @ jquery.js?1157:6089
append @ jquery.js?1157:6250
eval @ fintest-project.js?e33b:67

[PhoenixPhyre]

I think there is a bug in your new version - https://ibb.co/xHKLgMV



calculator.html:84 Uncaught TypeError: Util.getPresetOptions is not a function
at HTMLDocument.<anonymous> (calculator.html:84:18)
(anonymous) @ calculator.html:84

calculator.html:1 Uncaught (in promise) {message: 'A listener indicated an asynchronous response by r…age channel closed before a response was received'}message: "A listener indicated an asynchronous response by returning true, but the message channel closed before a response was received"[[Prototype]]: Object
Promise.then (async)
(anonymous) @ content_script_bundle.js:100
handleNewFeatures @ content_script_bundle.js:93
handleUpdatedNodes @ content_script_bundle.js:93
(anonymous) @ content_script_bundle.js:93
childList (async)
eval @ jquery.js?1157:6253
domManip @ jquery.js?1157:6089
append @ jquery.js?1157:6250
eval @ fintest-project.js?e33b:67


You may need to hard reload. The host caches things aggressively.

[Nefariis]

You may need to hard reload. The host caches things aggressively.

The cache clear fixed it - but I think your calculations might be still a little off.

<DELETED - I forgot you can only do flurry one time on the attack action, my bad>

[sambojin]

Not sure if it would be too fiddly, but want to put Brown Bear and Polar Bear on there, with some pretty poor assumed hit rates? +6 and +7 to-hit, but they don't improve. And probably -5% thereof every two levels or so after to account for enemy AC as they go up in CR quicker than your forms ever could. Just for some lvl2 and lvl6 Moon Druid numbers, to show why they drop off so consistently, even in their "damage forms". But start out pretty good, considering you didn't build for it.

Would be nice to include a 1 attack Summon Beastial Spirit (air) to show lvl3 casting, as well as the 2 attack at lvl7 and 3 attack at lvl11, assuming all ASIs are spent on Wisdom, to show summoning scaling as well, and show why they really don't drop off that badly, even with "the worst" summon. They don't have magic attacks (well, Shepherd's do at lvl6), but not everything has immunity or resistance either. Probably best to just assume 100% summon uptime and maximum slot used at each level, alongside +2 Wis at 4th and 8th level (druids are full casters, by about level 3-5 anyway). Wis20 by level8 isn't exactly what druids tend to do, but if they were going for raw to-hit/damage of summons, they could, and have plenty of other spells prepared alongside (even non-concentration ones). Makes it less "Moon'y", and more simply "Any Druid'y", for comparisons sake (I've made the comparison that Summon Fey Spirit is essentially just Summon Unoptimized Monk several times for instance). At least they don't get too much extra summon to-hit from spell to-hit from there-on afterwards in later tiers. They're maxed on Wis already, so those power spikes are accounted for.

I'll leave it to the document viewer to add those two lines together, to see where they sit in RED when being combat'y as a Moon Druid (and why restrain-on-hit or spell casting or better summons is probably better than your rather poor damage output at later levels, but even with the basics, you're not terrible. Well, maybe you are, the graph will tell).


((You can break it more by just casting a land spirit for more summon-HP, using a wildshape charge to have have a familiar ride around on it, so it gets auto-advantage through pack-tactics as it attacks, for more RED per cast/short-rest. With enough random BS, that tiny familiar has 3/4 obstruction/cover for any attacks against it as well. But that's well outside the bounds of this spreadsheet. I guess you as the PC probably should be doing stuff during at this time as well, being all Druid'y and stuff))

(((It's a nasty comparison to REDs, but when a lvl2 (upcastable) spell could quite capably add +0.5REDs or more to your character for potentially an hour, at any level of "character ability" at max, and you've got plenty of backfill to play with as well on "close enough", ouch. Oh no, you've maxed your casting stat early! That gives you more spell preps. Whatever else will you do? A hell of a lot more than a martial or skill-monkey, me-thinks)))

((((Remember, at lvl9-12, summoning several dragons spirits to ride around on throughout each day, is on the table for any Druid, summoning-wise. And no matter what RED says about that, that's a lot better. 10' reach and a choose-able resistance counts for a lot as an upgrade, on top of flying and large. The +1 breath attack is nice as well, but the least of it, in many ways. Not all druids get Elemental Form. But they all get a lvl9-10 "capstone spell selection", while still being able to do heaps, just as wizards get properly broken and fighters get really good. Which works great in all those up-to-lvl10/11 campaigns, because you had early strength all the way, and it's only at lvl7-11 that wizards truly start doing obscene things. But they weren't doing it for 1-2 spell preps, from lvl3 onwards))))

"A level 2-12 druid walks up to the bar.... And precisely what else were you expecting other than shenanigans on your system?"

I cast "Giant f'ing flying cat!" *can* be casted at every level of your silly little backstory, from lvl3 onwards. Choose your circle, by all means, but the facts stand for themselves. It's about as good as something not as good as that.

[PhoenixPhyre]

Not sure if it would be too fiddly, but want to put Brown Bear and Polar Bear on there, with some pretty poor assumed hit rates? +6 and +7 to-hit, but they don't improve. And probably -5% thereof every two levels or so after to account for enemy AC as they go up in CR quicker than your forms ever could. Just for some lvl2 and lvl6 Moon Druid numbers, to show why they drop off so consistently, even in their "damage forms". But start out pretty good, considering you didn't build for it.


I haven't done the other parts, but I did implement the "moon druid, bear forms only, produce flame at level 1" mode provisionally (probably needs more work). But the results were...lol. Best case (90% hit, 5% crit, 5% miss), the damage goes from 0.46 RED (throwing a single produce flame) at level 1 to 1.94 RED (brown bear starts at level 2) and then basically drops. Even polar bear form at level 6 is only a tiny jump (from 1.07 RED @ 5to 1.13 RED @6, where level 4 was 1.34 RED). At level 20, the polar bear is doing the same RED damage as a single produce flame was at level 1, 0.46 RED. Average best case is 0.82 RED. The curve is (sort of, at first glance, not rigorously) Poisson-shaped: rapid rise, followed by slower exponential fall-off.

Including accuracy makes this worse, dropping the average down to 0.75-ish (depending on the accuracy model) but doesn't really change the shape of the curve. Why isn't it more drastic on average? Because really, AC doesn't vary that much and partially washes out. Sure, your accuracy drops from 70% (level 2-3) down to 45% (level 17+) when measured against CR = level monsters[1]. But that's just drowned out by the massive non-scaling-damage issue. And since means are dominated by outliers, that really strong level 2 thing (which is also where you're most accurate) skews the math.

My guess (also based on experience) is that the expectation for moon druids is that their thing is
a) start concentration on some spell as action turn 1, especially things that don't take extra actions
b) shift into the biggest, beefiest form they have
c) use natural weapons as "cantrips" (aka at will) but their dominant source of contribution (damage, control, whatever) is coming from that long-duration spell.

This would also jibe with how druids have comparatively more concentration-based combat spells[2].

[1] where the default is "straight 65% except at level 9, where it's 70%", just as an easy reference point.
[2] even though they're middle-of-the-pack in total % of list that's concentration, oddly enough. But their combat-relevant spells, especially the good ones are heavily skewed to concentration.

[BRC]

I just want to say I used RED today! My PC's got a magic item as a reward, basically a flamethrower but for ice. It's cumbersome, and requires two hands at all times, basically replacing your offensive statblock when it's active, but giving you a whole new list of options. I anchored it around a "Cantrip" style ability using slightly-under RED for my PC's current level as my balance point.

[DarknessEternal]

I love your RED so much I've decided to throw a monkey wrench at you.

Is there a reverse-RED? That is, can you reverse the equations on monsters to see what the average expected incoming damage is in the same rounds/scenario?

[PhoenixPhyre]

I love your RED so much I've decided to throw a monkey wrench at you.

Is there a reverse-RED? That is, can you reverse the equations on monsters to see what the average expected incoming damage is in the same rounds/scenario?

Fairly simply. Only issue is the assumptions about fight length and monster selection. You could do individual monsters, or you could do DMG table monsters, or the median monster as implemented per CR.

[Malimar]

Basically, take the most dead simple rogue you can think of--a shortsword, always has sneak attack, uses their bonus action every turn for something else. Always hits, but never crits[1]. Starts at DEX = +3 and increases it at every opportunity until capped (level 8).
Assume a perfectly spherical rogue...

I don't have any actual complaints or objections to the project as described, it seems reasonable to me, I just wanted to make a joke.

[Frogreaver]

But that's the point. Multiple, not everyon.e Not everyone agrees monk sucks.

I've heard monk are not the best, but in my game the monk is holding their own with the ranger and warlock and they only just realized stunning fist isn't a bonus action last session. They haven't been using it until now because of that.

IMO, this observation tells us nothing without knowing how optimized those pc's are.

[Nefariis]

Would it be possible to somehow fit BB into the website?

Maybe three entries with 0%, 25%, 100% movement damage rider? Also maybe put it on a rogue (AT) and a non rogue as well?

I'm really really curious to see what BB is worth in terms of RED.

Also, what do you mean by "Always Power Attack"? is that Great Weapon Master?

[PhoenixPhyre]

Would it be possible to somehow fit BB into the website?

Maybe three entries with 0%, 25%, 100% movement damage rider? Also maybe put it on a rogue (AT) and a non rogue as well?

I'm really really curious to see what BB is worth in terms of RED.

Also, what do you mean by "Always Power Attack"? is that Great Weapon Master?

It's possible, certainly. I'll add it to the list of things to implement. I do want a better way of displaying these, but... Yeah. Not a designer.

And yes, that entry is GWM's -5/+10 option set to "always". That one is the same as the other gs champion but always attacking with a -5 penalty to hit and +10 damage per hit.

[Nefariis]

It's possible, certainly. I'll add it to the list of things to implement. I do want a better way of displaying these, but... Yeah. Not a designer.

And yes, that entry is GWM's -5/+10 option set to "always". That one is the same as the other gs champion but always attacking with a -5 penalty to hit and +10 damage per hit.

I'm not going to lie, I'm pretty shocked at that RED score - I definitely wasn't expecting that. Would you mind creating one more profile using PAM and GWM? I'm curious what the bonus attack with GWM adds to it.

[PhoenixPhyre]

I'm not going to lie, I'm pretty shocked at that RED score - I definitely wasn't expecting that. Would you mind creating one more profile using PAM and GWM? I'm curious what the bonus attack with GWM adds to it.

PAM is a bit obnoxious to implement. But it's on the list. Question about that? Do we care about order of feats? Ie GWM at 1, pam at 4 or vice versa?

[Nefariis]

PAM is a bit obnoxious to implement. But it's on the list. Question about that? Do we care about order of feats? Ie GWM at 1, pam at 4 or vice versa?

Haha, I purposefully omitted that in case you had a preference.

I think the standard build is PAM at 1 for the benefits and then GWM at 4 when your to-hit becomes a little better? but I have no preference, I think most people will be interested in the levels after 4 anyways.

[Gignere]

@OP this has been a good concept. Really helps crystallizes and makes it easier to compare DPR benchmarks. I’ve been toying around with different builds and seeing how many x Red they are doing, I’m assuming if you can get to or close to 3x Red the build is pretty over indexed for damage.

[PhoenixPhyre]

@OP this has been a good concept. Really helps crystallizes and makes it easier to compare DPR benchmarks. I’ve been toying around with different builds and seeing how many x Red they are doing, I’m assuming if you can get to or close to 3x Red the build is pretty over indexed for damage.

Yeah. 3.0 RED (especially average, but even a point instance) would be substantially above what seems to be the system's expected curve of ~1.5-ish (+- 0.25 or so) RED. Now, of course, you might have tables for which Nuclear Wizards (et al) are the norm. But if a team is routinely doing over 3.0 RED per person, the DM will have to substantially adjust combats to retain any sense of challenge.

Note that you could, under some situations (and with some tables) be fine if one person is doing 3.0+ RED--it seems that the system expects that all told, a party of 4 is doing roughly 4-ish RED. Two DPR-focused people at ~1.5 RED, plus 2 non-DPR focused people at ~0.5 RED (ie a control wizard or a healing cleric). So if you have one at 3 and the rest aren't doing much more than cantrip-level damage, then the system works fine. Party dynamics have to accommodate this, but in a party where most people don't care about combat, it can work fine.

[PhoenixPhyre]

Haha, I purposefully omitted that in case you had a preference.

I think the standard build is PAM at 1 for the benefits and then GWM at 4 when your to-hit becomes a little better? but I have no preference, I think most people will be interested in the levels after 4 anyways.

Preliminary calculations:

In the no-accuracy scenario, GWM with a GS from level 1 runs away early, then drops, leaving GWM/PAM (d10 weapon) ahead 2.56 to 2.28 in the end. In a more reasonable accuracy scenario, GWM/PAM is always ahead, but the gap narrows with levels (because you don't get more PAM attacks when you Action Surge, so it's basically static). Worst-case for accuracy (boss scenario), GWF/PAM spikes to a high of 2.46 @ level 5, but then drops a bit and stays roughly steady around 2.0, for an average of 1.94 RED vs the GWM/GS average of 1.66 RED. This is to be compared with the baseline GS Champion fighter at 1.43 RED average.

So on average, GWM adds ~0.23 RED while PAM (but shifting to level 4 for GWM) adds ~0.28 RED. As expected, inverting the levels at which you take the feats is a (small) decrease in overall overall damage (1.90 vs 1.94 RED in boss mode), driven entirely by dropping your damage at levels 1-3.

For the record, PAM alone boosts damage-over-RED (relative to a greatsword without PAM/GWM) before about level 11, after which it's virtually identical, leading to an average of 1.56 RED, for a final table of (relative to boss AC):



Type
Average RED


GWF style only, GS
1.43


GWF/Glaive + PAM@1
1.56


GWF/GS + GWM@1
1.66


GWF/Glaive + PAM@1 + GWM@4
1.94

[PhoenixPhyre]

Side note: for anyone code savvy, I went ahead and MIT-licensed a git repository with the calculation code at https://github.com/bentomhall/red-calculator. Pull requests welcome--the readme has some info about implementing new classes. Fair warning, the code's a bit of a mess.

Also, I fixed the fact that I wasn't taking into account fighters extra ASI at 6, as well as the fact that taking GWM at 4 means no ASI that level. Small changes in the overall numbers for the fighter-based presets.

And pushed the new stuff to the hosting site.

[animorte]

I meant to comment a week ago or so but didn't think about it. I think this is an excellent observation and design. It calculates a base line DPR to expect/build around. Observing the potential of damage output that you can easily account for whether you're a caster, basic martial, min-maxed, optimized, theme-built, what-have-you...

The work is appreciated, very easy to comprehend. AND it's my favorite color!

[Nefariis]

can you explain the four variations of accuracy again?

[PhoenixPhyre]

can you explain the four variations of accuracy again?

AC and Dex save bonuses are taken from the DMG table and my monster stats spreadsheet (respectively). I'm planning to allow you to switch between DMG tables and the (more representative) as-implemented numbers, but that's not there yet.

Default: no matter what, 90% of the time you hit, 5% of the time you miss, 5% of the time you crit. Unless you have an expanded crit range, which increases crit and decreases hit appropriately. Does not account for things like -5/+10.

"CR = level + 3" is what I think of as "boss mode". For each level, the AC/dex save bonuses are those of a monster whose CR is the level + 3 (roughly a "boss solo" fight). Adjustments from things like GWM are taken into account.

"CR = level / 2" is what I think of as "normal mode". Here, the comparison is level / 2, rounded down (so CR 1/2 at level 1, CR 10 at level 20), which comes from a calculation of the median expected multi-monster fight from Xanathar's guidance (assuming homogenous encounters).

"CR = level" is what it says on the tin.

But honestly--the differences between the latter 3 are really small. Because, as it turns out, hit chance is basically constant with the assumptions about modifiers. You can see this best by changing the mode to % hit and CR = level--it's a straight line at 65% except at level 9, where it's 70%. I'd be willing to bet that's how they derived the DMG numbers--assumed 65% hit chance with a +3 until level 4, +4 until level 8, and then +5 after that modifier. It's too unnatural otherwise.

[sambojin]

I haven't done the other parts, but I did implement the "moon druid, bear forms only, produce flame at level 1" mode provisionally (probably needs more work). But the results were...lol. Best case (90% hit, 5% crit, 5% miss), the damage goes from 0.46 RED (throwing a single produce flame) at level 1 to 1.94 RED (brown bear starts at level 2) and then basically drops. Even polar bear form at level 6 is only a tiny jump (from 1.07 RED @ 5to 1.13 RED @6, where level 4 was 1.34 RED). At level 20, the polar bear is doing the same RED damage as a single produce flame was at level 1, 0.46 RED. Average best case is 0.82 RED. The curve is (sort of, at first glance, not rigorously) Poisson-shaped: rapid rise, followed by slower exponential fall-off.

Including accuracy makes this worse, dropping the average down to 0.75-ish (depending on the accuracy model) but doesn't really change the shape of the curve. Why isn't it more drastic on average? Because really, AC doesn't vary that much and partially washes out. Sure, your accuracy drops from 70% (level 2-3) down to 45% (level 17+) when measured against CR = level monsters[1]. But that's just drowned out by the massive non-scaling-damage issue. And since means are dominated by outliers, that really strong level 2 thing (which is also where you're most accurate) skews the math.

My guess (also based on experience) is that the expectation for moon druids is that their thing is
a) start concentration on some spell as action turn 1, especially things that don't take extra actions
b) shift into the biggest, beefiest form they have
c) use natural weapons as "cantrips" (aka at will) but their dominant source of contribution (damage, control, whatever) is coming from that long-duration spell.

This would also jibe with how druids have comparatively more concentration-based combat spells[2].

[1] where the default is "straight 65% except at level 9, where it's 70%", just as an easy reference point.
[2] even though they're middle-of-the-pack in total % of list that's concentration, oddly enough. But their combat-relevant spells, especially the good ones are heavily skewed to concentration.


Thanks for that. Was about what I expected, just charted and graphed :)

Still think the Summon Beast comparison might be worthwhile. Just a bit fiddly to plug in.

+5to-hit, d8+6 damage at level 3 (16Wis start)
+6to-hit, d8+6dam at lvl4 (+2Wis ASI)
+7to-hit, d8+7dam at lvl5 (+PB, 3rd lvl slot)
+7to-hit, 2x d8+8dam at lvl7 (4th lvl slot)
+8to-hit, 2x d8+8dam at lvl8 (+2Wis ASI)
+9to-hit, 2x d8+9dam at lvl9 (+PB, 5th lvl slot)
+9to-hit, 3x d8+10dam at lvl11 (6th lvl slot)

It's just a worse case scenario of stuff "you could do", but won't, because there's better spells. Add cantrips or instants to taste.

They do scale pretty strangely, those summons do, especially compared to RED.

[PhoenixPhyre]

New version pushed--the new feature is a first pass at allowing the user to input custom data and graph it alongside other data.

The format is JSON array format: [number, number, number, ...]. The initial [ and final ] are critical--otherwise it will not do what you want. Data should be the raw accuracy-adjusted DPR for level i - 1 (0 indexed, so the first one is level 1). You can skip a level by inserting either 0 or null; that will produce a break in the line graph. If you don't provide all 20 levels, it will take it from the first and stop when it runs out of input. It will only be graphed if you select the 'Custom Data' preset. And changes to it won't do anything until you change the settings (preset, accuracy mode, output value). Accuracy for that preset is just whatever the RED accuracy is.

Note: It [I]will not try to accuracy-adjust for you. The RED baseline it uses as the denominator will be accuracy adjusted via the preset, so be careful.

[x3n0n]

Looking at SnB and GS Fighter (no Action Surge), they both have big bumps at 19th level and no additional bump at 20th level.

Does the code have the final Extra Attack (3) one level early?

Edit: yes, I found it, unless I've somehow cached an old version of presets.js:



attacks(level) {
if (level < 5) { return 1;}
else if (level < 11) { return 2;}
else if (level < 19) { return 3;}
return 4;
}


Darn off-by-one.

[PhoenixPhyre]

Looking at SnB and GS Fighter (no Action Surge), they both have big bumps at 19th level and no additional bump at 20th level.

Does the code have the final Extra Attack (3) one level early?

Edit: yes, I found it, unless I've somehow cached an old version of presets.js:



attacks(level) {
if (level < 5) { return 1;}
else if (level < 11) { return 2;}
else if (level < 19) { return 3;}
return 4;
}


Darn off-by-one.

Oops. No, that's a dumb on my part. Pull requests very much accepted if you have a GitHub account. Or I'll fix it when I can.

[x3n0n]

Oops. No, that's a dumb on my part. Pull requests very much accepted if you have a GitHub account. Or I'll fix it when I can.

I have an account and attempted to make a pull request. Let me know at some point if it didn't work. :) (Long-time coder, infrequent GH committer.)

[PhoenixPhyre]

I have an account and attempted to make a pull request. Let me know at some point if it didn't work. :) (Long-time coder, infrequent GH committer.)

PR merged and merged into my next overhaul (including webpack to modularize it a bit, which involves breaking that presets.js file into a bunch of smaller files per class), as well as deployed. The host is...aggressive...about caching, so it may take a while for it to actually show up right.