Should Might stay multiplicative or return to additive?

[KDubya]

I vote for multiplicative Might as it allows for more creativity with damage builds.  I like building characters with a ton of different modifiers and getting an impressive result.  

 

But multiplicative Might does not make builds more creative at all. It basically makes you max out your Might every time, not a whole lot of variety there.

 

The old additive system had Might roughly equal to Dex and Perception so you could focus on something other than Might and be as or near as effective.

 

Sure you don't have to max out Might when it is multiplicative but it is such a mechanical advantage that when designing for efficiency you'd be self-gimping to not max it out.

[Dr. Hieronymous Alloy]

@fiddlesticks - you're comparing the wrong values. The important value is not the Action Duration (the time it takes to perform an action), but rather the Action Rate* (number of actions performed per second). This value scales exactly as you would expect i.e. going from Dexterity 10 to Dexterity 20 increases your Action Rate by 30%, and going from Dexterity 10 to Dexterity 30 increases it by 60%.

 

To show this let's assume, for simplicity, that an action takes 1 second. Then at Dexterity 10 you perform that action once per second. At Dexterity 20 that action takes 1/1.3 seconds and hence you perform it 1.3/1 = 1.3 times per second. At Dexterity 30 that action takes 1/1.6 seconds and hence you perform it 1.6/1 = 1.6 times per second.

 

Thus the absolute value of each point of Dexterity is the same i.e. an increase in Action Rate of 3%. Of course the relative value of +1 Dexterity does indeed diminish e.g. 1.3 is 30% more than 1, but 1.6 is only 23.1% more than 1.3.

 

Actually the story isn't quite that simple thanks to the delay portion of an action cycle. You see, each action cycle is made up of several parts: the action itself, the recovery, sometimes the reload, and the delay. The first three of those are affected by Dexterity exactly as described above, however the delay is always 5 frames long no matter what your Dexterity is. This does in fact cause the absolute value of additional points of Dexterity to diminish, although if you actually run the numbers for realistic PoE values (Dexterity between 2 and 30) the difference is fairly minimal.

 

*In Pillars of Eternity this is called Attack Speed, although it applies to actions other than attacks and doesn't involve distance travelled so I prefer Action Rate.

 

EDIT: as a side note, if the bonus to Action Duration (not Action Rate) from additional points of Dexterity didn't diminish then you would, with high enough Dexterity, reach negative Action Duration. Hopefully it should be clear that this would be unwanted behaviour.

 

 

Ok, so am I right in thinking then that multiplicative Might and multiplicative Dex are roughly equivalent in terms of damage bonus given, except that

 

1) Might will have a slight edge in damage totals due to Dex's irreducible frames,

 

2) Dex will be better with on-hit effects because they'll proc more often,

 

3) Dex will also speed up non-damage actions (such as casting a crowd control spell).

 

?

[dunehunter]

 

@fiddlesticks - you're comparing the wrong values. The important value is not the Action Duration (the time it takes to perform an action), but rather the Action Rate* (number of actions performed per second). This value scales exactly as you would expect i.e. going from Dexterity 10 to Dexterity 20 increases your Action Rate by 30%, and going from Dexterity 10 to Dexterity 30 increases it by 60%.

 

To show this let's assume, for simplicity, that an action takes 1 second. Then at Dexterity 10 you perform that action once per second. At Dexterity 20 that action takes 1/1.3 seconds and hence you perform it 1.3/1 = 1.3 times per second. At Dexterity 30 that action takes 1/1.6 seconds and hence you perform it 1.6/1 = 1.6 times per second.

 

Thus the absolute value of each point of Dexterity is the same i.e. an increase in Action Rate of 3%. Of course the relative value of +1 Dexterity does indeed diminish e.g. 1.3 is 30% more than 1, but 1.6 is only 23.1% more than 1.3.

 

Actually the story isn't quite that simple thanks to the delay portion of an action cycle. You see, each action cycle is made up of several parts: the action itself, the recovery, sometimes the reload, and the delay. The first three of those are affected by Dexterity exactly as described above, however the delay is always 5 frames long no matter what your Dexterity is. This does in fact cause the absolute value of additional points of Dexterity to diminish, although if you actually run the numbers for realistic PoE values (Dexterity between 2 and 30) the difference is fairly minimal.

 

*In Pillars of Eternity this is called Attack Speed, although it applies to actions other than attacks and doesn't involve distance travelled so I prefer Action Rate.

 

EDIT: as a side note, if the bonus to Action Duration (not Action Rate) from additional points of Dexterity didn't diminish then you would, with high enough Dexterity, reach negative Action Duration. Hopefully it should be clear that this would be unwanted behaviour.

 

 

Ok, so am I right in thinking then that multiplicative Might and multiplicative Dex are roughly equivalent in terms of damage bonus given, except that

 

1) Might will have a slight edge in damage totals due to Dex's irreducible frames,

 

2) Dex will be better with on-hit effects because they'll proc more often,

 

3) Dex will also speed up non-damage actions (such as casting a crowd control spell).

 

?

 

 

Well, there are builds that doesn't rely on Dex, like Backstab, riposte and retaliation.

 

My point is any build who doing damage will benefit from Might, but they don't necessarily benefit from Dex.

Edited December 4, 2017 by dunehunter

[Dr. Hieronymous Alloy]

 

 

 

Well, there are builds that doesn't rely on Dex, like Backstab, riposte and retaliation.

 

 

 

Right.

 

But basically, with multiplicative Might (or Strength/Resolve), it's a build-dependent choice as to which is better; for most damage-focused builds,they're roughly equivalent, and even interchangeable.

 

In some ways it seems like it matters a lot less in Deadfire than in PoE, because in the original game, you needed a certain amount of Might to help punch through DR. Now with the move to Penetration/AR, it seems like they're basically interchangeable, especially the longer the battle lasts, unless your build has something specific in it that makes you prefer one or the other.

 

Counter-intuitively, I also suspect that the move to replace Might with Strength/Resolve actually makes Dexterity the best damage stat for casters. Hear me out: Casters are going to use their weapons a fair bit regardless, and Dex will help with both weapon and spell casting, and will help with non-damage spells also.

 

 

This also seems like a strong argument for multiplicative Might: if Might isn't multiplicative, Dex is the clearly superior choice, mathematically.

Edited December 4, 2017 by Dr. Hieronymous Alloy

[dunehunter]

Talking about delay in action cycle, I think every movement, every time being interrupted or being CCed, these can all be considered as some kind of delay between action. I think saying every point of Dex boost 3% of damage is in ideal situation. 

 

This is specially true for melee characters who will more likely to move around battlefield, and get CCed. They get less benefit from Dex than ranged characters because the actual delay between each action is higher than we thought.

Edited December 4, 2017 by dunehunter

[KDubya]

 

 

 

 

Well, there are builds that doesn't rely on Dex, like Backstab, riposte and retaliation.

 

 

 

Right.

 

But basically, with multiplicative Might (or Strength/Resolve), it's a build-dependent choice as to which is better; for most damage-focused builds,they're roughly equivalent, and even interchangeable.

 

In some ways it seems like it matters a lot less in Deadfire than in PoE, because in the original game, you needed a certain amount of Might to help punch through DR. Now with the move to Penetration/AR, it seems like they're basically interchangeable, especially the longer the battle lasts, unless your build has something specific in it that makes you prefer one or the other.

 

Counter-intuitively, I also suspect that the move to replace Might with Strength/Resolve actually makes Dexterity the best damage stat for casters. Hear me out: Casters are going to use their weapons a fair bit regardless, and Dex will help with both weapon and spell casting, and will help with non-damage spells also.

 

 

This also seems like a strong argument for multiplicative Might: if Might isn't multiplicative, Dex is the clearly superior choice, mathematically.

 

 

But in PoE where Might was additive, Dexterity was not a clearly superior choice. There Might was less essential the more damage modifiers you had, so for Ciphers and Rogues you had real completion between Might, Dex and Perception. It was only for spell casters trying to do a lot of spell damage and for Monks using fists (which had an increasing base damage but no damage modifiers like weapons had) that were really stuck with max Might everytime type builds.

 

A 26 (21 base + 5 from buff or rage) Might Helwalker with 10 wounds has a 36 Might which is +78% damage additively or 1.78X multiplicative. Consider a two handed sword base damage 23 with +100% damage adds does 64 damage in an additive system and 82 damage in a multiplicative. To get to the same damage in an additive system would require 62 Might.

[JerekKruger]

Ok, so am I right in thinking then that multiplicative Might and multiplicative Dex are roughly equivalent in terms of damage bonus given, except that

 

Under PoE's rule set no: due to the way the old damage reduction system worked, attacking with multiple weaker attacks was not as good as attacking with fewer strong attacks hence multiplicative Might would definitely have been better there.

 

In Deadfire it's less clear thanks to the changes to the armour/penetration system. Moreover if the action speed system is changed that might make a difference too.

[theBalthazar]

Obsidian want to dynamise his system.

 

You see that with the number of companion in party. You see that with penetration system. You see that with the multiplicative approach. You see that in their will to prohibit the slow mode. etc.

 

It is a knowingly choice...

 

This allows: to get rid of trash mobs faster. Clarify and accelerate the fight.

 

I would not be surprised if they want to position themselves against the approach of Original Sin 2 (turn-based) to attract a different type of player.

Edited December 4, 2017 by theBalthazar

[Dr. Hieronymous Alloy]

 

Ok, so am I right in thinking then that multiplicative Might and multiplicative Dex are roughly equivalent in terms of damage bonus given, except that

 

Under PoE's rule set no: due to the way the old damage reduction system worked, attacking with multiple weaker attacks was not as good as attacking with fewer strong attacks hence multiplicative Might would definitely have been better there.

 

In Deadfire it's less clear thanks to the changes to the armour/penetration system. Moreover if the action speed system is changed that might make a difference too.

 

 

 

Right, that's where I'm going with this argument, basically.

 

Under the old system, additive Might had a few clear advantages over (inherently multiplicative) Dex: it helped punch through DR, and it was "front loaded". Now, with AR/Pen, the "punch through DR" function of Might is gone; effectively, might was nerfed, at least in that regard. 

 

It seems like new multiplicative might serves to balance it out vs. multiplicative Dex. Different character builds will have reasons to pick one or the other, but for a lot of builds, they'll be interchangeable, especially given the longer fight durations that seem to be a design goal in Deadfire. 

 

It is true that some class combinations can stack Might very high and get very high damage totals now but that seems to me to be more a function of the ready availability of different stacking Might buffs than it does of the background numbers. 

 

 

What would be really interesting would be to see someone calculate the approximate DPS value of Perception under the new system. Without grazing it seemed like a must-buy but with grazing coming back and misses less likely, I suspect that the mathematical value of a point of Perception is going to be less than the value of a point of Might or a point of Dex, and really they should be approximately equal. 

 

Conversely, the important thing now isn't a comparison between what Might does now and what it did in the last game, but between what Might does now and what the other damage stats (Dex and Per) do in *this* game. Is multiplicative might better or worse than Dex or Per *in Deadfire*? That's the relevant question.

 

[i *suspect* the graph for might will be linear relationship between points and dps, while the graphs for dex and per will show diminishing returns, but that may not be a big deal depending on the shape of the curves and whether or not the intersection points are above or below the stat totals that can be actually achieved in game).

Edited December 4, 2017 by Dr. Hieronymous Alloy

[Stormhammer117]

 

 

 

 

Well, there are builds that doesn't rely on Dex, like Backstab, riposte and retaliation.

 

 

 

Right.

 

But basically, with multiplicative Might (or Strength/Resolve), it's a build-dependent choice as to which is better; for most damage-focused builds,they're roughly equivalent, and even interchangeable.

 

In some ways it seems like it matters a lot less in Deadfire than in PoE, because in the original game, you needed a certain amount of Might to help punch through DR. Now with the move to Penetration/AR, it seems like they're basically interchangeable, especially the longer the battle lasts, unless your build has something specific in it that makes you prefer one or the other.

 

Counter-intuitively, I also suspect that the move to replace Might with Strength/Resolve actually makes Dexterity the best damage stat for casters. Hear me out: Casters are going to use their weapons a fair bit regardless, and Dex will help with both weapon and spell casting, and will help with non-damage spells also.

 

 

This also seems like a strong argument for multiplicative Might: if Might isn't multiplicative, Dex is the clearly superior choice, mathematically.

 

 

Also, since spell damage AND deflection is going to Resolve, only certain types of weapon-based offensive characters will favor Might.  If it is not multiplicative, this type of character will not be very competitive compared to a caster or a hyrbid.  Might should have SOME reason to dump points into it - if it's not a very good force multiplier then what is the point?  

[KDubya]

 

 

 

 

 

Well, there are builds that doesn't rely on Dex, like Backstab, riposte and retaliation.

 

 

 

Right.

 

But basically, with multiplicative Might (or Strength/Resolve), it's a build-dependent choice as to which is better; for most damage-focused builds,they're roughly equivalent, and even interchangeable.

 

In some ways it seems like it matters a lot less in Deadfire than in PoE, because in the original game, you needed a certain amount of Might to help punch through DR. Now with the move to Penetration/AR, it seems like they're basically interchangeable, especially the longer the battle lasts, unless your build has something specific in it that makes you prefer one or the other.

 

Counter-intuitively, I also suspect that the move to replace Might with Strength/Resolve actually makes Dexterity the best damage stat for casters. Hear me out: Casters are going to use their weapons a fair bit regardless, and Dex will help with both weapon and spell casting, and will help with non-damage spells also.

 

 

This also seems like a strong argument for multiplicative Might: if Might isn't multiplicative, Dex is the clearly superior choice, mathematically.

 

 

Also, since spell damage AND deflection is going to Resolve, only certain types of weapon-based offensive characters will favor Might.  If it is not multiplicative, this type of character will not be very competitive compared to a caster or a hyrbid.  Might should have SOME reason to dump points into it - if it's not a very good force multiplier then what is the point?  

 

 

Might was additive in PoE and was useful for anyone wanting to do damage. It just wasn't the no brainer max it that multiplicative Might is.

 

Mathematically, a multiplicative Might is a much better investment than Dexterity or Perception for a weapon using damage dealer. Not to mention that the more you have the better even more will get you.

 

An additive Might will still be useful but  one could choose to keep Might a little lower and instead raise Dex or Perception and still be as viable. Sure enough the game is easy enough that one does not need to min max to be viable, but when one choice of maxing a multiplicative Might is so much better from an efficiency standpoint than anything else that you could do then it takes away from being able to be creative with the builds and things start to get really cookie-cutterish.

[peddroelm]

High Dex[haste] offers a bit of 'protection' towards interruptions[literal stun lock if attacked by multiple fast attackers when wielding sufficiently slow weapon + high recovery combo ] and even CC .. Ability to react in time to changing battlefield conditions .. [once you've started a slow attack .. you've forfeited control of your character for the duration of the swing + recovery - you might have an ability to counter some nasty stuff you notice coming your way - you might not get the change to 'cast' it in time .. ]

 

Shorter cool downs [even for Auto Attacks] provides tactical flexibility .. Becomes painfully apparent in solo situations .  

Edited December 5, 2017 by peddroelm

[hilfazer]

 

Mathematically, a multiplicative Might is a much better investment than Dexterity 

 

Is Dex additive with other action speed bonuses?

[Dr. Hieronymous Alloy]

 

 

Mathematically, a multiplicative Might is a much better investment than Dexterity 

 

Is Dex additive with other action speed bonuses?

 

 

 

I've been trying to figure out the formula for "how much of a DPS boost does each successive point of Dex give" and it's  . . . complicated, even after some very patient people took their time to explain it to me. . As I understand it, all the action speed bonuses stack, but "multiplicative" and "additive" aren't either exactly right because the more you reduce action speed, the closer you approach a theoretical minimum action speed, so it's a curving function, not a linear one.

 

 

 

 

Mathematically, a multiplicative Might is a much better investment than Dexterity or Perception for a weapon using damage dealer. Not to mention that the more you have the better even more will get you.

 

 

 

 

Yeah, this is correct, but it's a limited statement. Multiplicative Might / Str  is the best investment for a pure weapon damage character; Might / Res is the best investment for a pure casting damage character. 

 

For hybrid characters (i.e., weapon and spell damage, or damage and CC), Dex is probably the best offensive stat investment, then either a split between Str & Res or Perception, with the math on that changing depending on things like what other Accuracy / Dex / Per buffs are available and how the Dex curve shakes out. 

 

Given the proposed upcoming graze range of 25-49, and the critical hit damage boost of +25%, as best I can figure out, the damage boost from each additional point of Per is, roughly, [copying from someone else's math here, slightly modified; any mistakes are mine] 

 

50 + (25/2) == 62.5% expected damage on a normal attack vs. equivalent defense, adjusted for accuracy / miss / graze rate

 

50+ (25/2) + (1*1.25) == 63.75% with an additional point of accuracy (as one miss shifts up to a graze, graze shifts up to a hit, hit shifts up to a crit). 

 

So going by that, each additional point of Per will give a roughly 1% damage boost (while also helping spell accuracy, etc.) 

 

Once you get up to an accuracy boost over +25 though, if there are lots of Perception or Accuracy buffs available, this of course changes dramatically; changing a hit to a crit is far less dramatic in effect than eliminating a miss. On PoTD or against higher level enemies Per would be more valuable, again because of the importance of reducing the miss rate. 

 

Unfortunately I'm not savvy enough to figure out -- or at least I haven't yet -- where the break-point is where additional points of Dex become less valuable than additional points of  Str, Res, and/or Per, presuming everything else (weapon, damage, attack speed, opponent's defense, etc.) is being held equal.

 

If the above math is correct, though, it does seem to indicate that a hybrid damage character is better off alternating points in Str and Res, rather than stacking Per; half of a 3%-per-point bonus being preferable to the whole 1% bonus. Per would only really be useful for CC or crit-seeking builds.

Edited December 5, 2017 by Dr. Hieronymous Alloy

[JerekKruger]

I've been trying to figure out the formula for "how much of a DPS boost does each successive point of Dex give" and it's  . . . complicated, even after some very patient people took their time to explain it to me. . As I understand it, all the action speed bonuses stack, but "multiplicative" and "additive" aren't either exactly right because the more you reduce action speed, the closer you approach a theoretical minimum action speed, so it's a curving function, not a linear one.

 

It's not a curving function. The Action Cycle Duration (the time it takes for a full Action Cycle to occur) decreases with Dexterity in a non-linear fashion, but that's not what's important for DPS. What's important for DPS is the Action Rate (the number of Actions per second), which happens to be inversely proportional to the Action Cycle Duration.

 

Think of it this way: suppose a character with Dexterity 10 takes exactly 1 second to Attack and Recover (let's ignore the 5 frame delay for now). Now suppose that character increases their Dexterity to 43. They now take 1/(1+0.99) ~ 0.503 seconds to Attack and Recovery, which means that their Attack Rate has increased from 1 Attack per second to 1.99 Attacks per second. All else being equal this results in an increase of 99% to their DPS i.e. the increase to DPS is linearly proportional to the bonus to Attack Speed given by Dexterity.

 

The fact it's multiplicative was determined through testing and through looking at the code, with MaxQuest being the one who worked it out fully in this thread: https://forums.obsidian.net/topic/86684-mechanics-the-big-attack-speed-conundrum/

 

TL;dr: each point of Dexterity increases the number of times you attack per second by 3% (compared to Dexterity 10) and hence increases your DPS by 3%.

Edited December 5, 2017 by JerekKruger

[Dr. Hieronymous Alloy]

 

 

The fact it's multiplicative was determined through testing and through looking at the code, with MaxQuest being the one who worked it out fully in this thread: https://forums.obsidian.net/topic/86684-mechanics-the-big-attack-speed-conundrum/

 

TL;dr: each point of Dexterity increases the number of times you attack per second by 3% (compared to Dexterity 10) and hence increases your DPS by 3%.

 

 

 

Yeah, I was looking at the chart in this post : https://forums.obsidian.net/topic/86684-mechanics-the-big-attack-speed-conundrum/?p=1808225

 

Maybe I'm misreading it, but it looks like Dex is multiplicative but there is a diminishing-returns effect that occurs (partly due to the irreducible frames, partly because you're taking bites out of a progressively smaller pie). 

 

Going by that chart, again if I'm reading it correctly, going from 10 to 11 dex results in a DPS increase of 2.3%, whereas going from 20 to 21 dex gives 1.7%. So, if those numbers are correct and I'm understanding them correctly, then Dex is always inferior to multiplicative Might for damage based characters, always superior to a point of either multiplicative str/ or multiplicative res for hybrid damage dealers, and always superior to Per for everyone  -- at least for values attainable in the character creator. 

 

The real story seems to be that Per just dramatically underperforms for everybody. 

Edited December 5, 2017 by Dr. Hieronymous Alloy

[Dr. Hieronymous Alloy]

edit: wait, actually, my above math on Per is wrong (I think):

 

Each point of per adding 1.25% (below +25 accuracy above target's defense) means that each point of Per is actually worth a 2% or so dps boost:

 

62.5% (expected damage on a given swing when accuracy equals defenses, given graze and miss damage adjustment); increase of 1.25% to 63.75%, from one additional point of accuracy, is 2% of 62.5%; so, each point of Per gives a 2% damage increase, so long as you don't have too many other accuracy bonuses.  

 

That's a better rate of dps-per-stat-point than a high Dex gives, but worse than a low dex gives. So if you have dex buffs or gear you may be better off with Per, but at character creation and otherwise naked, Dex is probably a better buy at least for the first few points.

 

I apologize if I'm turning this thread into my scratch math sheet -- my desire for #'s sometimes exceeds my grasp of #'s!

 

Unfortunately I don't think any of that really gets us too much closer to answering the original question re: multiplicative v. additive might. A straight 3% multiplicative might is definitely better than any other stat for pure damage dealers, but a straight 3% additive might might be inferior.

 

What about if might/res/str gave a 2% multiplicative bonus, instead of 3% ? That would be more in line with the bonuses from Dex and Per, with the limitation that it was damage only (not other actions) but the benefit that it was linear and not subject to diminishing returns. It seems like at least part of the issue isn't so much additive vs. multiplicative, as that Might (or even str/res) gives a much stronger DPS bonus than other stats do. 

Edited December 5, 2017 by Dr. Hieronymous Alloy

[dunehunter]

edit: wait, actually, my above math on Per is wrong (I think):

 

Each point of per adding 1.25% (below +25 accuracy above target's defense) means that each point of Per is actually worth a 2% or so dps boost:

 

62.5% (expected damage on a given swing when accuracy equals defenses, given graze and miss damage adjustment); increase of 1.25% to 63.75%, from one additional point of accuracy, is 2% of 62.5%; so, each point of Per gives a 2% damage increase, so long as you don't have too many other accuracy bonuses.  

 

That's a better rate of dps-per-stat-point than a high Dex gives, but worse than a low dex gives. So if you have dex buffs or gear you may be better off with Per, but at character creation and otherwise naked, Dex is probably a better buy at least for the first few points.

 

I apologize if I'm turning this thread into my scratch math sheet -- my desire for #'s sometimes exceeds my grasp of #'s!

 

Unfortunately I don't think any of that really gets us too much closer to answering the original question re: multiplicative v. additive might. A straight 3% multiplicative might is definitely better than any other stat for pure damage dealers, but a straight 3% additive might might be inferior.

 

What about if might/res/str gave a 2% multiplicative bonus, instead of 3% ? That would be more in line with the bonuses from Dex and Per, with the limitation that it was damage only (not other actions) but the benefit that it was linear and not subject to diminishing returns. It seems like at least part of the issue isn't so much additive vs. multiplicative, as that Might (or even str/res) gives a much stronger DPS bonus than other stats do. 

 

Yeah as I posted earlier here, I'm Ok that Might is a multiplier if it has diminishing effect as Dex does

 

But still I like it as an addictive because it frees cipher and rogue to choose other stats but also provide those who doesn't access to large basic damage bonus a way to boost their damage. Because if it's an addictive, the larger basic damage bonus u have, the less gain u get from Might.

Edited December 5, 2017 by dunehunter

[JerekKruger]

@Hieronymous - ah I see what you mean. Part of it is, as you say, the irreducible delay frames (this is why the increase from Dexterity 10 to 11 doesn't give a 3% increase). The other factor is that MaxQuest is giving the relative rather than absolute increase in DPS.

 

The absolute increase would be the percentage increase in DPS compared to Dexterity 10; the relative increase is the increase in DPS going from Dexterity X to Dexterity X+1 and this does go down as X increases. The absolute increase gained from Dexterity 21 is 29.34%, not that far below the stated value of 33% (and the difference here is entirely due to the 5 frame delay).

 

Note that there are a similar diminishing relative returns from Might in Deadfire. To see this suppose a character doors 100 damage with Might 10. Then with Might 11 they'll do 103 damage i.e. a 3% increase. At Might 20 they'll do 130 damage and at Might 21 they'll do 133 damage, so the relative increase from going from Might 20 to Might 21 is 133/130 = 1.0231 i.e. 2.31%, less than 3%.

 

TL;dr: Multiplicative Might does have a diminishing return in relative increase to DPS in exactly the same way Dexterity does.

Edited December 5, 2017 by JerekKruger

[Dr. Hieronymous Alloy]

@Hieronymous - ah I see what you mean. Part of it is, as you say, the irreducible delay frames (this is why the increase from Dexterity 10 to 11 doesn't give a 3% increase). The other factor is that MaxQuest is giving the relative rather than absolute increase in DPS.

 

The absolute increase would be the percentage increase in DPS compared to Dexterity 10; the relative increase is the increase in DPS going from Dexterity X to Dexterity X+1 and this does go down as X increases. The absolute increase gained from Dexterity 21 is 29.34%, not that far below the stated value of 33% (and the difference here is entirely due to the 5 frame delay).

 

Note that there are a similar diminishing relative returns from Might in Deadfire. To see this suppose a character doors 100 damage with Might 10. Then with Might 11 they'll do 103 damage i.e. a 3% increase. At Might 20 they'll do 130 damage and at Might 21 they'll do 133 damage, so the relative increase from going from Might 20 to Might 21 is 133/130 = 1.0231 i.e. 2.31%, less than 3%.

 

TL;dr: Multiplicative Might does have a diminishing return in relative increase to DPS in exactly the same way Dexterity does.

 

 

 

Ahhhhhh ok, now I see what you're saying. 

 

SO really what I want is a chart of the relative value of each additional point of Might, Dex, and Per, and also Str / Res, from 3 to about 40 or so (since that's generally the max practical ceiling and if you can get your stats over that this kind of fiddling with numbers no longer matters anyway). 

 

And what such a chart will probably show is that even with multiplicative Might, each additional point of Might gives somewhere between a 3% and 2% bonus, each additional point of Dex gives a similar but slightly smaller bonus, and each point of Per gives at most a 2% bonus (with a threshold at about +25 accuracy where it drops considerably). 

Edited December 5, 2017 by Dr. Hieronymous Alloy

[JerekKruger]

And what such a chart will probably show is that even with multiplicative Might, each additional point of Might gives somewhere between a 3% and 2% bonus, each additional point of Dex gives a similar but slightly smaller bonus

 

Essentially yes. The increase from Might 39 to Might 40 is 1.6%, and the increase from Might 2 to Might 3 is 3.95% (yep, you actually get the biggest relative increase going from Might 2 to Might 3), and Dexterity will be a little less.

 

However the irreducible delay makes the Dexterity calculation more complicated as the relative increase changes depending on what other +Attack Speed bonuses a character has (the same isn't true with Might and other +Damage bonuses). The closer your Recovery is to zero, the less relative increase a point of Dexterity gives (essentially because that irreducible 5 frames becomes a bigger proportion of what remains). This is why in PoE, characters who could reach 0 recovery usually left their Dexterity at 10.

 

... and each point of Per gives at most a 2% bonus (with a threshold at about +25 accuracy where it drops considerably).

 

Perception seems to me to be a lot more complicated to assign a value to since it depends on things like +Crit Damage bonuses, Miss>Graze>Hit>Crit modifiers and, of course, the difference between your base Accuracy and the enemy's Deflection.

Edited December 5, 2017 by JerekKruger

[Dr. Hieronymous Alloy]

 

And what such a chart will probably show is that even with multiplicative Might, each additional point of Might gives somewhere between a 3% and 2% bonus, each additional point of Dex gives a similar but slightly smaller bonus

 

Essentially yes. The increase from Might 39 to Might 40 is 1.6%, and the increase from Might 2 to Might 3 is 3.95% (yep, you actually get the biggest relative increase going from Might 2 to Might 3), and Dexterity will be a little less.

 

However the irreducible delay makes the Dexterity calculation more complicated as the relative increase changes depending on what other +Attack Speed bonuses a character has (the same isn't true with Might and other +Damage bonuses). The closer your Recovery is to zero, the less relative increase a point of Dexterity gives (essentially because that irreducible 5 frames becomes a bigger proportion of what remains). This is why in PoE, characters who could reach 0 recovery usually left their Dexterity at 10.

 

 

 

 

Ok, that makes sense. GIven what we can see in the beta, then, it seems like multiplicative might and multiplicative dex are roughly equivalent now -- given that Might/Str/Res buffs only damage while Dex boosts everything and makes on hit/crit effects pop more often -- and whether or not they're equivalent in the full game will depend on how many +action speed effects there are in the full game, which we can't know yet.

 

EDIT: It does mean that Dex is clearly superior to Strength / Resolve for anyone who ever plans to use a weapon and a damage causing spell, both.

 

 

... and each point of Per gives at most a 2% bonus (with a threshold at about +25 accuracy where it drops considerably).

 

 

Perception seems to me to be a lot more complicated to assign a value to since it depends on things like +Crit Damage bonuses, Miss>Graze>Hit>Crit modifiers and, of course, the difference between your base Accuracy and the enemy's Deflection.

 

 

 

Yes. . .  but . . . most of those potential modifiers to Perception make the relative value of additional points worse. Any increase in accuracy past +25, you're shifting grazes to crits, rather than misses to crits. The same is true for any graze to hit or hit to crit effects. The most common critical hit damage modifier I've seen in the beta is "blunted criticals" on guns (why! so cruel!). 

 

Upshot is that most of the time, if you're choosing between Per and either Dex or Might, Per is the worst choice for DPS. 

 

Even on Path of the Damned   .. presuming it keeps the same +15 to all defenses penalty. . .  that just means that, presuming otherwise equivalent attack/defense, you're going to miss on a 0-40, graze on a 41-65, hit on a 66-115, and crit above that. So:

 

expected value of a normal attack on PotD: 35 + (25/2) = 47.5

With one additional point of Perception:  36 + (25/2) =  48.5

 

So that works out to still roughly a 2% damage increase per point ( a little more, but not enough more to make up for Crits being unreachable; it's a rounding error). 

 

The only real modifier that would make Per better would be either more accuracy (but if you went up to 2 acc per point that would be too much) or bonus damage on a critical hit (but then you run into a feedback loop issue where you have increasing rather than diminishing returns).

Edited December 6, 2017 by Dr. Hieronymous Alloy

[MaxQuest]

Have been avoiding posting in this thread for awhile as could not choose between the variants asked in the starting post.

Both approaches have their pros and cons, neither being perfect. In the end have arrived to conclusion that there can be a middle ground, and this mixed approach appeals most to me.

 

So:

 

- First of all let's take a look at DEX contribution to our auto-attacking dps:

 

The absolute gain from 1 extra point in DEX is the same. But relative gain keeps going down due to intrinsic returns.

But the important part are the values, which will be used for comparisons below

 

- Now let's take a look at MIG contribution to our auto-attacking dps:

 

 

Multiplicative MIG seems to look as a more desired approach, because it's relative gain:

- does not change on crits/grazes and does not dance around

- is decoupled from weapon quality, and does not get diluted by it

- is equal and consistent with relative gain of DEX

 

Also we can take a look at relative dps increase from PER (around the acc-def = 0):

 

 

As we can see, additive MIG and especially additive grazes and crits do make relative gain from PER heavily fluctuate. And that that doesn't happend if those are multiplicative.

 

 

 

On the other hand if we would have multiplicative MIG it would hurt builds that deal phys damage while having low MIG score, such as cc-ciphers and deathblow rogues. What gives?:

 

 

TL.DR. Am suggesting a mixed v4 solution to the might problem from the top post.

I'm aware of the idea to beta test the new STR-RES in the next update. And those could use this as well.

 

---

P.S. In case one questions, why Deadfire does not use additive MIG like PoE1 did: it is because of lack of DR.

Vs zero/low-DR targets MIG impact is way lower than that of DEX, when it comes to weapon-based damage. Mainly because it gets heavily diluted by other effects. But the existence of high-DR targets could skyrocket the relative gain of 1 MIG point, e.g:

 

If all dmg coefficients stayed additive in Deadfire, MIG effect would be too low for weapon based damage.

Edited January 14, 2018 by MaxQuest

[Dr. Hieronymous Alloy]

See, that's what comes of crunching the numbers. Solid work. 

 

It could be a little confusing but I suspect it wouldn't matter too much because anybody sufficiently willing to dive into the numbers to the point that they started to care about things like 'is the bonus from Might additive or multiplicative?" could figure that proposed system out.

 

Simple phrasing is "bonuses of the same type are additive, then you multiply each group of bonuses." 

Edited December 17, 2017 by Dr. Hieronymous Alloy

[MaxQuest]

Simple phrasing is "bonuses of the same type are additive, then you multiply each group of bonuses."

Pretty much it.

Although it rises the question of how to tell that effects in question are of the same type)

 

---

Btw, have taken a look a bit at the current auto-attacking action cycle. The spotted differences from PoE1 are:

- damage is inflicted at the end of attack duration, not somewhere in the middle as before. This makes it almost impossible to 'abuse' Quick Switch like we could by switching weapons right after shooting and thus eliminating both reloading AND recovery.

- I haven't noticed the inter-action delay. After recovery end, the attacking animation was starting immediately. Still can't assert that there is no delay at all, since could achieve only ~26 fps during frapsing instead of constant 30.