How to Dragoon Per Second: A DPS Paradigm

[lootbox]

Hi Dervy,

Thanks for all the great information, especially all the number crunching! I saw your reddit post on how crit chance and crit damage scale with the crit stat, and had some thoughts.

It seems the crit stat weight is an approximation based on the fact that crit doesn't scale linearly. Some expected value analysis shows that crit scales quadratically (since it has a linear impact both on crit chance and crit damage).

I wanted to try to figure out how that compares to str but I'm a little confused about your formulas for weapon skills and auto-attacks:

Weapon Skills:
((Potency/100)*(WD/25+1)*(STR/9)*(DET/7290+1)*BUFFS)-1.5

Auto-Attacks DPS:
((AA_DMG/33.04+1)*(STR/6.92)*(DET/6715+1)*(1/AA_DELAY)BUFFS)-1.5

What do those values represent? Is it raw DPS? If so, where does the -1.5 come from? In terms of raw damage numbers, I would think 1.5 is negligible.

[Dervy]

Which value? Do you specifically mean the -1.5? That's a value to counteract two things:

1) Potency doesn't scale at the value of the tooltip. My theory is that Each Point of Potency is actually a 0.01005 multiplier of damage, rather than it being Potency/100. So if a skill is 360 Potency, instead of it being a 3.6 multiplier, it's actually a 3.618 multiplier.
2) To accommodate for Rounding Error Values until I have enough data. So far, those damage models are the most accurate I've ever gotten them to be.

It was Viridiana's idea to convert it back to Potency/100, then resolve for the final value, otherwise, it would be (Potency*0.01005), and -1 at the end.

The Weapon Skill formulae is the Raw Damage at X Potency. If you wanted to calculate your DPS, instead of using the Potency of a skill, you would use the Potency Per Second (PPS) value of your rotation.

Auto-Attacks are the same. If you wanted to calculate DPS of Auto-Attacks, you'd use that formulae. If you wanted to calculate the raw damage per Auto-Attack, you would remove the (1/AA_DELAY).

If you're asking about the Division Coefficients, those are just scalers. Instead of multiplying it by a Decimal, you can inverse the operation by dividing it by a whole number.

[lootbox]

snip

Gotcha. Here's what I came up with:

Say a non-critical hit deals D damage (substitute in your formula for weaponskill, auto-attack, or DoT damage), and c = crit - 354. Then:

Expected Damage = D * non-crit chance + D * crit chance * crit damage multiplier
Expected Damage = D * (1 - crit chance) + D * crit chance * crit damage multiplier

crit chance = 0.000233 * c + 0.049511233
crit damage multiplier = 0.000233 * c + 1.4493610

...wolfram alpha magic...

Expected Damage = (5.4289e-8 * c^2 + 0.000116237 * c + 1.02225) * D

Take the derivative with respect to c:
(1.08578e-7 * c + 0.000116237) * D

This gives a formula for how much we should expect damage to increase for each additional point of crit, given the current crit value (where, again, c = crit - 354). At higher values of crit, each additional point increases damage by more.

I don't know if this is useful at all to you but I was bored at work and figured I'd share.

EDIT: I think WA did some weird rounding first time through.
Updated formula which includes critical chance buff here.

[Dervy]

No, this would help me out heck load actually. I was manually changing Critical Hit Rating... Rotation PPS... typing it out the data... to check the effective damage per Critical Hit Rating. This will speed up a lot of things for me, thank you so much.

[kyrios91]

So i get the chance to parse on a dummy (No party, food, and pot buffs) and I hit 1055 DPS in 3 mins. I wonder if that's a good standard?

[seekified]

So i get the chance to parse on a dummy (No party, food, and pot buffs) and I hit 1055 DPS in 3 mins. I wonder if that's a good standard?

Yeah, you don't need to worry. You can probably get a little higher I reckon but at that point it's down to optimising GK timing and making sure that time spent moving for 4ths is kept to a minimum. I'd try again and keep going to about five minutes though as that'll let you do your entire opener again after 240 seconds, which should boost numbers a bit I imagine.

On a different note, I recall Yoshi saying they were looking at the animations for jumps to make them even more fluid, but I certainly don't feel a difference from 2.5. Kinda get the feeling that several changes are being held until they have data from Alexander.

[lootbox]

Viridiana's post reminded me that I forgot to account for critical chance buffs. Same thing, but including critical chance buff b:

Expected Damage = D * non-crit chance + D * crit chance * crit damage multiplier
Expected Damage = D * (1 - crit chance) + D * crit chance * crit damage multiplier

crit chance = 0.000233 * c + 0.049511233 + b
crit damage multiplier = 0.000233 * c + 1.4493610

...wolfram alpha magic...

Expected Damage = (5.4289e-8 * c^2 + (0.000116237 + 0.000233 * b) * c + 1.02225 + 0.449361 * b) * D

Take the derivative with respect to c:
(1.08578e-7 * c + 0.000116237 + 0.000233 * b) * D

Changes bolded all involving b. If Internal Release is on, b = 0.1. If Battle Litany is on, b = 0.15. If both are up, b = 0.25. If nothing is up, b = 0 and you get the same formula from before.

Hope this helps!

[SunnyHirose]

I feel like they merely made the animation locks shorter again, and were more talking about overcoming some technical barriers or something.

Like, I can feel the difference as a long-time DRG main. And don't get me wrong, I could tell that the 2.4 changes made jumps during T11 Nerve Gas non-lethal. The dives now feel more like Shoulder Tackle. But as long as it does lock you and doesn't hop over AoEs, it'll always be risky, and therefore make no serious difference to how DRG is played.

[seekified]

I feel like they merely made the animation locks shorter again, and were more talking about overcoming some technical barriers or something.

Like, I can feel the difference as a long-time DRG main. And don't get me wrong, I could tell that the 2.4 changes made jumps during T11 Nerve Gas non-lethal. The dives now feel more like Shoulder Tackle. But as long as it does lock you and doesn't hop over AoEs, it'll always be risky, and therefore make no serious difference to how DRG is played.

Yeah, you're right. Guess I played far too much WAR towards the end of 2.5, but the change this time around is definitely smaller than the 2.4 stuff.

Am I right in thinking that Elusive Jump instantly tells the game that you're in the destination position when it's triggered? I've always felt that Spineshatter/DFD move you through the field whereas EJ is more of a teleport with a nice animation, but maybe it's just fast.

[Aiurily]

regarding jumps, i do feel like SSD/DFD have a shorter lock before you can move at least. maybe that's just me though. i can do FT/CT into SSD/DFD and hit the 4th positional pretty easily because of that. technically you can use jump after FT/CT and still hit the 4th too; but the timing is weird, and you have to react fast. the process would go like this.

TT>WT/F&C proc>you move into position as soon as you see the proc>hit the jump button after you are in position>hit WT/F&C

this doesn't clip your GCD at all if you do it right, since there's about a 0.6s window after a GCD where you can't do an oGCD anyway, you abuse that timeframe to position yourself for the positional to eb used as soon as you recover from jump animation lock. of course, the realistic time to execute the movement is probably close to 0.3s~ because you have to receive the proc and process it into your mind which direction to go.

and then with geirskogul usage, after you get used to it, you find patterns that allow you to get double during BfB pretty easily after a while. generally i find it safe to just build up to 30s of BoTD (by hitting 4th on 15s~ duration on it) and then just judge from there how often i need to use it to run dry by the time BoTD is back up, and how it'll aline with my CDs.