D&D General Why DPR Sucks: Discussing Whiteroom Theorycrafting

[FrogReaver]

But I think even the DPR assumptions are off. For instance, a melee fighter may have better DPR than a ranged fighter, but the melee fighter loses DPR whenever they aren't in range of their opponent. Meanwhile the ranged fighter's DPR is consistent for most of the range.

Range isn't clearly better than melee though. As if you want to factor in range you also need to factor in whatever contributions the melee fighter is giving the party by going into melee (say keeping the enemy away from those with concentration spells going). That is, when it comes to party play - a lack of front line characters tends to be a problem, even though ranged are easier to argue in favor of.

DPR doesn't factor the most important factors of how a fight will pan out. A paladin that's in the thick of it may have amazing DPR and falls behind the bard because they died in round 2 for the whole 6 round combat while the bard got an extra 4 rounds tinkering with their animate objects from a safe distance.

Sure, or the bard would have been dead in round 1 and lost his concentration immediately without the paladin there at all.

Meanwhile, concentration is expected to always be on, but a cleric whose Spirit Guardian's drop early now has much lower DPR since they have to spend time re-applying it. If they even have the spellslots left.

Sure, and a fighter who gets disadvantage does as well.

DPR isn't normally about factoring in every potentiality. It's about computing what you are going to be able to do in a typical adventuring day (or round - depending on your focus).

 

[Asisreo]

Range isn't clearly better than melee though. As if you want to factor in range you also need to factor in whatever contributions the melee fighter is giving the party by going into melee (say keeping the enemy away from those with concentration spells going). That is, when it comes to party play - a lack of front line characters tends to be a problem, even though ranged are easier to argue in favor of.



Sure, or the bard would have been dead in round 1 and lost his concentration immediately without the paladin there at all.

This is exactly why DPR is flawed. What I said could happen, what you said could happen, but DPR factors none of that in. If the ranged enemies are being forced into melee, their DPR probably tanks since they start using their actions to escape.

A bard's DPR is assumed to be round-by-round concentration always on, but if they get concentration knocked out of them, they may have had better DPR concentrating on something more defensive and using a different option.

DPR isn't normally about factoring in every potentiality. It's about computing what you are going to be able to do in a typical adventuring day (or round - depending on your focus).

But the less potentiality you compute with, the muddier of a picture you're presenting.

If the fighter always goes down first (which would be typical of a frontliner), then DPR only tells the beginning portion.

Meanwhile, a life cleric focusing on DPR loses out on HPR (Healing per Round) which may be more potent and a more efficient use of their abilities.

 

[FrogReaver]

This is exactly why DPR is flawed. What I said could happen, what you said could happen, but DPR factors none of that in. If the ranged enemies are being forced into melee, their DPR probably tanks since they start using their actions to escape.

A bard's DPR is assumed to be round-by-round concentration always on, but if they get concentration knocked out of them, they may have had better DPR concentrating on something more defensive and using a different option.


But the less potentiality you compute with, the muddier of a picture you're presenting.

If the fighter always goes down first (which would be typical of a frontliner), then DPR only tells the beginning portion.

Meanwhile, a life cleric focusing on DPR loses out on HPR (Healing per Round) which may be more potent and a more efficient use of their abilities.

Arguing something isn't perfect and therefore it is bad is not a good argument. Something can be imperfect and still be useful.

 

[Asisreo]

Arguing something isn't perfect and therefore it is bad is not a good argument. Something can be imperfect and still be useful.

I'm not arguing it's imperfect. I'm arguing that it's insufficient and misleading.

Data collected by live sessions is imperfect but it can provide sufficient information. The information is conclusive and it includes things like mobility, AoE's, average DM'ing styles, etc. If there's enough data, it could be conclusive by a much tighter margin.

But if I were to tell you "Yes, we ran the numbers on the rollercoaster and we calculated it to be very safe...but we haven't gotten around to live testing it yet. Wanna try it out for a ride?" Are you just going to say "sure! Numbers are better than live tests anyways"?

 

[Maxperson]

Arguing something isn't perfect and therefore it is bad is not a good argument. Something can be imperfect and still be useful.

cough alignment. cough

 

[Maxperson]

A few years back I GMed a campaign and decided to track some data for the group. It was a combat heavy campaign. The players could expect 3 or more fights in a session. Here is the aggregate for the 5 characters that appeared more than 10 times.

Player	Damage	Kills	Blood Loss	Knocked Out	Team Killer	Sessions
Hunter	2062	40	528	3	85	18
Open Hand	2119	34	834	5	32	22
Evoker	3842	63	176	0	103	21
Goliath Bard grappler	455	4	171	0	27	15
Devotion paladin	3124	43	1025	4	57	22

(Team killer is damage dealt to teammates, the rest should be self-explanatory)
This data covers from level 1 to 11 and this campaign was pre-Xanathar.
Excess damage was not counted. If an enemy had 1 hit point then your crit counted for 1 damage.

This is just a snapshot of one table during one campaign but I thought it illustrated nicely how each player contributes in different ways. You've got the numbers above but here's what I saw as a GM:

1. The Evoker is clearly captured in the numbers. She brought the pain and did everything she could to avoid it herself.
2. The Paladin is mostly captured in the numbers. High hit points, shield master feat, and aura meant he was the tankiest person around. He would charge in and smite the heck out of anyone and everyone. The only thing missing is the healing he did.
3. The monk was a hill dwarf whose Dex and Wis never went above 16. Definitely not built or utilized in the way most people would expect for a monk. And yet, he dealt decent damage, put plenty baddies out of their misery, stunned targets, and tanked a lot of damage with only one more KO than the paladin. All of that and he was often the only short rest reliant member of the group.
4. The Hunter was the best at finishing off enemies at range (strong damage to kill ratio). He also brought some healing, was the scout of the group, and would off-tank as well. Just an all around great contributor.
5. Then you have the grapple bard. Not great numbers right? Doesn't matter, he was the most encounter destroying thing to have ever sat at my table. The grapples, the control spells, the healing, and the flood of inspirations were devastating. But that doesn't show up at all.

I'm with the OP when it comes to how much I value things like DPR. I find it useful but it'll never tell the whole story for a game like DnD.

One things brought up here which may have already been mentioned(I skipped a lot of posts) is how classes interact with one another. The Bard was an encounter destroyer in part because of how he helped out the others. How much of a rogues DPR is due to another party member allowing him to flank and gain advantage? How much damage is done, because the PC is conscious and/or able to act due to making a save he would have missed without the Paladin's aura? And so on.

White room DPR just simply fails, because the game isn't played in a white room.

 

[NotAYakk]

A few years back I GMed a campaign and decided to track some data for the group. It was a combat heavy campaign. The players could expect 3 or more fights in a session. Here is the aggregate for the 5 characters that appeared more than 10 times.

Player	Damage	Kills	Blood Loss	Knocked Out	Team Killer	Sessions
Hunter	2062	40	528	3	85	18
Open Hand	2119	34	834	5	32	22
Evoker	3842	63	176	0	103	21
Goliath Bard grappler	455	4	171	0	27	15
Devotion paladin	3124	43	1025	4	57	22

(Team killer is damage dealt to teammates, the rest should be self-explanatory)
This data covers from level 1 to 11 and this campaign was pre-Xanathar.
Excess damage was not counted. If an enemy had 1 hit point then your crit counted for 1 damage.

This is just a snapshot of one table during one campaign but I thought it illustrated nicely how each player contributes in different ways. You've got the numbers above but here's what I saw as a GM:

1. The Evoker is clearly captured in the numbers. She brought the pain and did everything she could to avoid it herself.
2. The Paladin is mostly captured in the numbers. High hit points, shield master feat, and aura meant he was the tankiest person around. He would charge in and smite the heck out of anyone and everyone. The only thing missing is the healing he did.
3. The monk was a hill dwarf whose Dex and Wis never went above 16. Definitely not built or utilized in the way most people would expect for a monk. And yet, he dealt decent damage, put plenty baddies out of their misery, stunned targets, and tanked a lot of damage with only one more KO than the paladin. All of that and he was often the only short rest reliant member of the group.
4. The Hunter was the best at finishing off enemies at range (strong damage to kill ratio). He also brought some healing, was the scout of the group, and would off-tank as well. Just an all around great contributor.
5. Then you have the grapple bard. Not great numbers right? Doesn't matter, he was the most encounter destroying thing to have ever sat at my table. The grapples, the control spells, the healing, and the flood of inspirations were devastating. But that doesn't show up at all.

I'm with the OP when it comes to how much I value things like DPR. I find it useful but it'll never tell the whole story for a game like DnD.

Awesome data.

My conclusions are different.

DPR is a measure. If you measure up poorly, it doesn't mean you suck. You can easily contribute elsewhere.

But from your numbers?

A hunter ranger is widely considered the 2nd least effective class-subclass in 5e. Beastmaster ranger being the bottom of the totem pole.

Evoker and Paladin put out 50% more pain each.

50% is big. Like, 2 monk/rangers could be replaced with 2 paladins.

Even moreso, Paladin is better at spike damage than sustained; while both monks and rangers have more endurance. So that Paladin damage was probably bigger when they needed it most, and less when it wasn't important.

What more? Paladins have crazy high charop ceiling. Unless you mentioned it, I presume this was a baseline paladin, nothing janky.

---

Monks have nearly no charop room. They fare poorly compared to rogues, especially once you leave "total damage if everything hits" basic anaylsis and account for misses and disengages and the like.

And Rogues also suffer from a low charop ceiling. But a higher one than a monk. Booming blade, elven accuracy crit fishing, shadowblade, martialmweapon dip/feat + scimitar of speed ready trick, mainlining haste potions, ring of spell storing haste (+ready trick); piles of stuff that give 30% to 80% DPR boost.

That amount of charop lets a rogue keep up with a GWM barbarian champion crit fisher (9 barb/11 champ) or a battlemaster sharpshooter or a sorcladin or a gloom stalker assassin fighter alpha stike or a sorcerer ape army or ... piles of stuff.

And monks? 20%-30% from the same level of effort that gave the rogue up to 80%.

With no feats and MCing, the monk is PHB ranger-tier damage. It hasnone strong trick - stun storm.

I suspect part of the problem is that the subclasses with features at 11 don't carry their weight. In theory a level 11 subclass feature can match the DPR boosts of primary class level 11 features, but in practice they don't.

Gloom 11 is head and sholders above hunter or beast 11, but still sucks compared to fighter 11. Monks 11 same. Rogues get a non-combat boost at 11.

Use DPR with increasing fidelity as one measure. The issue is that Monks don't measure up in DPR, falling 33% or more behind stronger classes, and don't seem to bring enough other stuff to the table to make up for that gap.

 

[Asisreo]

A hunter ranger is widely considered the 2nd least effective class-subclass in 5e. Beastmaster ranger being the bottom of the totem pole.

I think this is exactly the point in which the community decides DPR isn't the complete picture for some reason.

If I were to pull up an optimal Beastmaster Ranger's DPR, it would fair just fine.

Gloom 11 is head and sholders above hunter or beast 11, but still sucks compared to fighter 11. Monks 11 same. Rogues get a non-combat boost at 11.

Again, from a pure DPS standpoint, the beastmaster does very well at this tier.

Let's go through the motions.

A beastmaster Ranger that chooses a Giant Poisonous Snake as the companion can have a +8 to-hit and does 1d4+6+3d6= 19 damage in one turn. That's 20 damage in a single round, no setup. If the target saves, the damage is 10.75 which is practically par for the course DPR with a higher to-hit.

At level 3, the gloomstalker (assuming archery) has a +7 to-hit that does 2d8+6+2d6 = 22 if all attacks connect. The problem is that every round afterward does 1d8+1d6+3 = 11 damage on average. The .25 difference is probably made better by the fact that it does 19 on a failed save, even if the saving throw is low. The beastmaster can also use their bonus action spells for something else like setting up a snaring strike on their next turn.

At level 4, the gloomstalker gets stronger with their initial to-hit of +8 and damage of 2d8+2d6+8 = 24 if all attacks connect. They then do 1d8+1d6+4=12 damage on average.

The thing with the 4th-level beastmaster is that after the initial turn, they too can do the same damage as the gloomstalker, 1d8+1d6+4=12. Or they could keep going with the snake companion for the chance to do 19 damage at any given turn.

5th-level gets pretty crazy for the Beastmaster. Not sure if you notice, but the beastmaster is actually the only ranger subclass that DPR benefits directly from level 5 feature.

At 5th-level, the Gloomstalker's damage is 3d8+3d6+15= 39 with +9 to-hit on all attacks for the first round, going to 2d8+2d6+10=26 for the rest of the time.

At 5th-level, the beastmaster does 1d4+4d6+1d8+11=32 damage on a failed save and does 1d4+1d6+(3d6÷2)+1d8+11=26.75 damage on a failure, with +9 to-hit for both the snake and ranger. So now the consistent damage of the beastmaster has surpassed the consistent damage of the gloomstalker.

Another big boost to DPR at level 11, where the beastmaster does 2d4+26+2d8+2d6+6d6 =68 damage (if 2 saves are failed). With a +11 to-hit with your bow and +10 with the snake. If they fail 1 save, the damage is 2d4+26+2d8+2d6+(3d6÷2)+3d6=62.75. If they fail both saves, the damage is 57.5

Meanwhile, the gloomstalker's initial damage at 11th level is 3d8+15+3d6 = 39. The gloomstalker's continuous damage afterwards is only 2d8+10+2d6=26 damage for the rest of the rounds. (Disclaimer: the gloomstalker gets a re-roll on a miss, bringing their effective to-hit every round higher. It's harder to calculate without a given target, though. I will if requested.

Damage for beastmaster also gets a massive boost at level 15, due to share spell (and heightened prof bonus). They now do 2d4+28+2d8+4d6+6d6=79 if they fail 2 saves, does 2d4+28+2d8+4d6+(3d6÷2)+3d6=71.75 if they fail 1 save and now doing 2d4+28+2d8+4d6+(6d6÷2)=66.5 damage per round.

The gloomstalker still does exactly the same damage as level 11 with a bump to to-hit from the proficiency up.

So past level 11, beastmaster is far-and-away better than the gloomstalker in terms of pure DPS. The main force of damage is also magical, so there's no worries about that, either. Even if the enemies fail their saves (which they're likely to do), they still take far more damage than the gloomstalker. Much, much more.

Of course, it depends on the enemies's resistances and immunities for poison, but DPR doesn't account for that, does it? Besides, it could be that humanoids are the most common enemy, so poison resistances might just be rare enough to ignore, even at these higher levels.

Use DPR with increasing fidelity as one measure. The issue is that Monks don't measure up in DPR, falling 33% or more behind stronger classes, and don't seem to bring enough other stuff to the table to make up for that gap.

Yet the beastmaster past level 11 is pumping some extreme numbers, yet seen as the punching bag of the class options. Wonder why? Their DPR is so high, what could possibly go wrong?

 

[TaranTheWanderer]

Every time I look for advice on character options, I find Class Optimization threads and they are all about DPR. After reading, I leave thinking, "Crap, I think my halfling Barbarian that weilds a longsword in two hands concept is bad." (or whatever)

The problem, as you pointed out, is all the feats 'in blue' are combat feats and all the abilities 'in red' are the fluff abilities that actually make the game interesting to play the other 66.66% of the time. (I'd say our games are combat 1/3 of the time).

It would be better to have stats based on how many d20s get rolled to accomplish a task. Then weigh that based on the pillar.

In combat, there are piles of d20 rolled before you can defeat the challenge and, therefore, damage is important. Also, things like bardic inspiration and certain spells become important because hitting more often means rolling d20s less to win the fight. Each individual d20 roll holds very little weight since there are, probably, a hundred or more rolls in a combat.

The other pillars usually use less d20s to accomplish a specific task. A social conflict might require 1-5 rolls while an exploration conflict/task might require only a few survival checks. Any spell that takes away the need to roll a d20 (comprehend languages spell vs having to roll Intelligence, for example) can garner points based on the weight of the particular roll. Spells that boost a d20 roll are worth points based on how well they ensure success etc...

I think that would be a better way to decide the overall utility of a class.

 

[NotAYakk]

Asis, that is why I mentioned increased fidelity.

You modelled a 0 AC foe (all attacks hit) eith -infinity saves. You can improve that model.

Gloom stalker 3 model. Hand crossbow, XBE, 16 dex. Despite ease, not invisible. HM round 1.
4d6+6 (20) at +8 to hit.

On an 18 AC foe, this is .6*4.5+.55*20+.05*14 DPR, or 14.4, 11.7 later rounds.

Snake BM is +8 for 1d6+6 then save for 3d6+2 at DC 11 (for half). We'll model 18 AC and +1 con save.

.55*(9.5+.45*(12.5)+.55*.5*(12.5))+.05*3.5=11.25

Doesn't need to use hunter's mark, but lower DPR than Gloomstalker XBE. Also gloomstalker isn't in darkness, whichher damage out of the water.

How about level 5, gloom in darkness with SS against 18 AC?

Snake BM is +9 to hit for 1d6+7(10.5) save for 3d6+3(13.5), plus a +10 to hit 1d10+1d6+4(13) HM enhanced crossbow shot. 19ish DPR.

Gloom is +4 to hit for 6d6+39(60) with advantage, or +9 to hit at 6d6+9(30) without. On AC 18 hits 5/9th of the time and crits 1/10 for 35ish DPR in darkness, 20ish in light.

Level up to 11. Snake is +10 to hit at 2d6+16(23) save for 6d6+8(29) plus +11 to hit at 1d10+1d6+5 (14). 35ish DPR.

Gloom is +6 at 6d6+45 (66) reroll 1 miss, advantage. Chance of a miss is like 70%-80% (3 attacks, each with 1/3 chance of missing, has a 70% chance of a miss). So in darkness does (22*(.7)+7*.1)3.7+4.5(.8) or about 63 DPR. Sustained, as after first round can bonus action crossbow. Switching targets does cost extra DPR.

In light, +11 to hit for 3.7*2d6+5 (44ish). Hits on a 7 for 70% accuracy. (.7*(12)+.05*(7))*3.7 is 32ish DPR. A tad less than the Snake BM.

So in light, against a low Con decent AC foe, our Snake BM has similar DPR as our gloomstalker.

DPR isn't everything, so we start looking around. Snake damage is magical, but also poison. If magic weapons are rare this could help BM.

L 11 Snake BM has a 44 HP sponge with low AC that has nearly no HD and takes more than 8 hours to replace (need to find a giant poisonoua snake, then befriend it, then spend 8 hours).

Gloomstalker has invis-in-darkness features.

Gloomstalker scales better with published treasure. Give gloom magic weapon and bolts...

BM gains less from HM, and has bonus action mostly free.

BM has more feats free; above build uses none. GS uses 2.