Mid-level math?

It's interesting to me that so much math discussion centers around level 20 comparisons. This seems odd to me, because not only do most people spend muh longer playing below "level cap," but there are also a bevy of very-high-level abilities that drastically change the potential of various classes.

Meanwhile, Stalker0 has a thread on ENWorld claiming that a raging level 10 barbarian does about double the DPR of a fighter or monk. If that's accurate (and I'm at work and can't do the deeper checking to confirm), it seems like a huge problem.

Has anyone noticed issues like this in classes as they level?
I'll get around to it eventually. I think DPR from 5-15 is probably the most important range to calculate for actually.

P.S. After level 11 or so, most classes don't really see a significant increase in DPR...
I don't dispute Stalker0's numbers, but we used say that you can't draw a line with just one point. That comparison takes place on very favorable ground for the barbarian.

1) At 10th level the Barbarian gets the Unchecked Fury feature which is a modest DPR booster.
2) At 11th level the Fighter gets a third attack per round which is a significant DPR booster
3) Advantage boosts DPR the most when To Hit percentages are low, and the example target in this case was AC18, where you've seen mostly AC15 for people doing reference calculations.
4) The fighter he used wasn't optimized for damage (although to be fair on this point, his purpose was not simply damage output.) 
5) The Barbarian has 4 rages per day at that point and both DPR and survivability go way down without rage. The fighter has no such limitation. (Of course if you die to the 1st Fire Giant of the day, you don't really get to fight the 5th) 
Good points, yakuba, and I'll add the obvious - the raging barbarians doesn't get attacks of opportunity or other reactions, and can be knocked out of rage if he can't attack each turn. So the fighter has more tactical flexibility. But I'd still like to see more points on that line, because (for example) a class that sucks until level 12 or whatever has a bigger problem than a class with slightly subpar DPR at level 20 (when you're probably not in that many straightforward slugfests anyway, and DPR may well be secondary to other factors).
So... I actually think calculations at a few different levels would be beneficial. Say... 3,9 and oh... 12.
"What's stupid is when people decide that X is true - even when it is demonstrable untrue or 100% against what we've said - and run around complaining about that. That's just a breakdown of basic human reasoning." -Mike Mearls
I think that sounds about right, but I'd add level 6 and 15 to that mix as well. We need to do more than compare DPR though. We need to compare the proportion of expected damage to the average monster HP for monsters from those levels, and vice versa. The problem is, monster math is still a mess and class bonuses are probably getting a redesign next packet so I'm not sure the dev's are even going to consider this sort of feedback until that've even tried to address things.

Basically, we're a year and a half in and we haven't even begun to really playtest the real system yet.
why obseess with math?  game about roll play not math.  simple add and subtact is all you need to play.  anyone can do.  no need to worry bout how math works in game.  dm can fix any problems wuth monmsters to strong or not



That's really not true. Minor differences are to expected, and there will be holy wars over what counts as a minor difference. But you don't want the Barbarian being strictly better in combat then the fighter when both classes are stricly combat classes, that will kill fun for the people who want to play that strictly inferior fighter. Similarly, you don't want the rogue or the bard with all their out of combat coolness to also be strictly better than the fighter in combat for the same reason. Big disparities (as we currently see with the Monk) matter, and small differences keep message boards going.

See... because the packet math will befixed later, I'm not too concerned about disparity across all levels. 3 is for me (and others) the new 1. By level 9 every class will have 2 attacks. I think at 6 the fighter's DPR would have a serious jump due to getting his two attacks relatively early. 9 puts it a level after most classes have 2 attacks. 12 puts it a level after martial classes willhhave likely acquired more damaging abilities (divine strike or three attacks, for example). I'm more interested in knowing if there are broken or weird behaviors I should warn my PCs about. A few packets back the rogue sneak attack wasn't worth trading away advantage after certain levels of AC. Similarly, the level 1 cleric swift casting inflict wounds was stupid powerful. This is why a few levels across different classes would make for a useful comparison. It could also give us a better ability to eyeball numbers in the math-fixed packet.
"What's stupid is when people decide that X is true - even when it is demonstrable untrue or 100% against what we've said - and run around complaining about that. That's just a breakdown of basic human reasoning." -Mike Mearls
So... I actually think calculations at a few different levels would be beneficial. Say... 3,9 and oh... 12.



Here are some (mostly) at will DPR numbers at Levels 3,5,9,12 and 15. They seem mostly smooth, but that is deceptive, there are some very large level to level jumps, in particular for the Warrior and the Berserker. Most notable thing to me is how the Thief Damage falls off relative to the Fighters and the Berserker. It starts off that a good sneak attack is quite strong and by L15 it's pretty meh. I didn't use multi-attacks from the Dueling Master, Shield Master or Archery Master Feats as there wording is (at best) ambiguous as to how they affect multiple attacks per action.

I'll put the other assumptions below the table for those who might care. 

Key (only for the unobvious):
Archer: Fighter/Warrior 
Sniper: Rogue/Thief
Sniper Sn: Rogue/Thief with Sneak Attack
Sniper Sn+A: Rogue/Thief with Advantage


 



























































































































































BuildL3L5L9L12L15
Mage2.485.636.08912.83
Evoker7.437.6514.6119.03
Hasted14.8515.329.2338.05
Archer5.6513.4517.426.127.68
Knight4.7311.514.722.0523.48
Warrior6.615.4529.2743.7547.67
Berserker5.27.0817.0426.9532.21
Raging9.0811.2831.9844.7951.04
Sniper3.434.685.77.158
Sniper Sn5.188.8812.5316.1120.21
Sniper Sn+A7.6812.3616.8523.4228.11
Thief5.356.958.159.5511.25
Thier Sn7.9812.8317.3621.8426.18
Thief Sn+A11.1316.4121.2226.5232.96


Assumptions:
16 post-racial in primary stat. 
Berserker and Fighter use 'Power Attack' after attaining Heavy Weapon Master feat
vs AC15
Thief uses TWF
At L15 Thief uses Fencing Master additional attack with -5 on ALL attacks for both basic attack and Advantage attack but not Sneak Attack.

Weapons:
Archer: Long Bow
Knight: Long Sword (knowingly not optimum)
Warrior: Great Axe
Berserker: Great Axe
Sniper: Short Bow
Thief:L3-9 Shortsword/Shortsword  L12-15 Rapier/Shortsword

Feats:
Mage/Evoker: +2 Int/ +2 Int/Irrelevant

Archer: +2 Dex/ +2 Dex / Master Archer / Irrelevant / Irrelevant
Knight: +2 Str/ +2 Str/ Tactical / Heavy Armor Master / Shield Master
Warrior: +2 Str/ Heavy Weapon Master/ +2 Str / Heavy Armor Master/ +2 Con
Berserker: +2 Str/ Heavy Weapon Master/ +2 Str

Sniper: +2 Dex/ +2 Dex / Archery Master / Stealthy
Thief: +2 Dex / +2 Dex/ Duel Wielder / Fencing Master

I think a big problem with these math threads is you guys arent actually play testing these builds your adding in a vaccuam. When you add the aspect of taking damage wierd party combination, player choice, bad luck streaks. Party team work. It changes alot of these numbers. I understand this math needs to be done but actual adventuring needs to be the next step(and not at lvl 20, I mean who starts their guys there).

These new forums are terrible.

I misspell words on purpose too draw out grammer nazis.

I think a big problem with these math threads is you guys arent actually play testing these builds your adding in a vaccuam. When you add the aspect of taking damage wierd party combination, player choice, bad luck streaks. Party team work. It changes alot of these numbers. I understand this math needs to be done but actual adventuring needs to be the next step(and not at lvl 20, I mean who starts their guys there).



Actual adventuring would be incredibly inefficient for posters who are big on math; despite their prolific calculations and posts, it is unlikely that they actually play enough playtest sessions to test for every party combination and also to experience the corner cases in statistical probability.

Deviation from the average is an important statistic though, I can agree. It's good to know how far on any given round or in any given encounter that you can expect a character to deviate from its average performance (Standard Deviation is an important tool for assessing how often those "bad luck streaks" and other corner cases come up)

I notice that certain posters like to assume that evaluating a class includes adding the benefits of another class's buffing capabilities to it; this isn't a bad idea, but it should be noted that each class -should- function within its areas of expertise without relying on the presence of another class, or you're testing a class combination (not a bad idea) as opposed to a single class. 
Yea. My sense is that wizard cantrips scale up too slowly in mid levels, but then get too potent near the end of the game (due to the fact that they scale like nothing else). Also, they are too potent with haste after about level 10. They should make sure that their DPR stays about the same, but break it up into two attacks/cantrip casts, and make it max out (like it would at level 20) at level 8. That would keep it scaling by the same pattern as everyone else and keep the game balanced across more levels. Rogue's could use just a tiny damage bump in later level. Maybe around level 10 those d6's should become d8's? Otherwise, those numbers seem about right. 
Thanks for breaking that down, Yakuba (are you ever yakoob elsewhere online?).
I'm actually pretty pleased by those numbers. Even though a few are wicked jumps. The assassin is likely to have a slightly improved average DPR w/ its assassination, so the thief not hitting as hard is... well... good. I can see new subclasses evolving from there to handle different rogue play styles.
The hasted evoker is pretty intense. I'm not sure if a hasted battlemage shouldn't put out that level of damage. Granted, it maybe should be brought down due to a full-on hasted nova would use more destructive spells.
I like that the builds presented were reasonable and suboptimal. I think a party w/ this group wouldn't feel like any other PC was drastically outshining them.
I think damage bounds are more important than the St Dev for these calculations. In 4e, I had a monk and a damage-focused ranger in the same party. The monk had a tighter average, but when the ranger rolled high...
"What's stupid is when people decide that X is true - even when it is demonstrable untrue or 100% against what we've said - and run around complaining about that. That's just a breakdown of basic human reasoning." -Mike Mearls
People interested in math on this board generally aren't actually picking their favorite classes mathematically - they're trying to help establish a good mathematical balance between classes now, during the playtest, so that it's possible to just pick your class for RP reasons without feeling useless by the time we're playing "for real."
Thanks for breaking that down, Yakuba (are you ever yakoob elsewhere online?).



No. I only post here, and not that much.

Even though a few are wicked jumps. The assassin is likely to have a slightly improved average DPR w/ its assassination, so the thief not hitting as hard is... well... good. I can see new subclasses evolving from there to handle different rogue play styles.



I think an assassin will have lower overall damage compared to a Thief. Cunning action, and especially Cunning action + Mobile allow the thief to advantage a lot (especially halflings). They really need to a Thug or Rake concept that basically a scaled down fighter with Sneak attack. Not my idea of fun, but you have acknowledge the high demand for that kind of build.

I like the assassin concept but it goes from hard hitting to pretty sissy by level 12 or so, and if the assassin can't really do damage in round 1, it's not too fun. I love the Thief anyway. I switched from planning on playing a paladin to  playing a thief for this playtest period.

The hasted evoker is pretty intense. I'm not sure if a hasted battlemage shouldn't put out that level of damage. Granted, it maybe should be brought down due to a full-on hasted nova would use more destructive spells.

 
I'd really like to see a viable damage oriented arcane caster, though not necessarily the mage. I keep hoping the warlock will come back largely unchanged, I really liked that class.

That said, to me, the evoker is the most boring caster concept ever. It's kind of amazing that people are so fixated about what evokers can do. What I'd really like to do is go back to the very first packet where ray of frost did not damage and you could realistically build a completely non-damaging mage. Adding damage to ray of frost really killed that concept.