 |
06/07/07, 6:53 PM
|
#271
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
Oh, and an idea I had in the vein of valuation of stats, would be to create a tool that takes the same input as my mini-spreadsheet, and creates three curves for each shot, one that is damage by crit change starting from your current crit, one that is damage by rap change starting from your current rap, and one that is damage by rap and crit change starting from your current rap and crit (this last one would of course be a 3d graph). Each could have a given range (I'd think based on itemization points, maybe 10 item levels worth of points or something). It could also be extended to have a shot cycle as input and create a 3d graph for dps changes based on those two axis.
The reason I think this would be useful is because the single-increment valuation is useful for a gauge, but if you add more than that increment your valuation starts losing more and more accuracy (because for each variable the valuation actually depends on that variable, savvy?). I mean, while people dish out the idea that 14RAP = 1DPS the reality is that the RAP-DPS graph is a curve when taking fluctuating crit values into consideration (14RAP at x crit is really zDPS, but at y crit it is aDPS). With the curves you could get a feel for relative slopes and figure out at any give point when to shift your focus for itemization.
Does something like that sound useful to you guys? I kinda think so, since it would be gear-independent and allow for a better sense of what kind of upgrades are more beneficial (instead of what specific upgrade is best buy guess and checking it in one of the inclusive spreadsheets).
Update: I think I settled on the best visualization for it. A single 3d graph with DPS, Crit %(rating?), and RAP as the axises. It'd take the same inputs as my spreadsheet plus a shot cycle/duration and pop out the proper graph. For spice the agi curves for each of the buff-states could be drawn on as well. Maybe allow for the suspension of one axis (dropping it down to DPS and Crit for instance) and the isolation of the agi curves (so the axises would be DPS and agi). Hmm...yes, I think those would be interesting. Could a spreadsheet do that? Or would it need to be a 3d graphing tool?
Last edited by Groggan : 06/07/07 at 7:34 PM.
|
|
|
|
|
06/07/07, 7:47 PM
|
#272
|
|
Von Kaiser
Night Elf Hunter
Laughing Skull
|
Hey Cheeky Band of Eternity friendly/honored have the wrong stats for them.
|
|
|
|
|
06/07/07, 8:01 PM
|
#273
|
|
Don Flamenco
|

Originally Posted by Groggan
Oh, and an idea I had in the vein of valuation of stats, would be to create a tool that takes the same input as my mini-spreadsheet, and creates three curves for each shot, one that is damage by crit change starting from your current crit, one that is damage by rap change starting from your current rap, and one that is damage by rap and crit change starting from your current rap and crit (this last one would of course be a 3d graph). Each could have a given range (I'd think based on itemization points, maybe 10 item levels worth of points or something). It could also be extended to have a shot cycle as input and create a 3d graph for dps changes based on those two axis.
The reason I think this would be useful is because the single-increment valuation is useful for a gauge, but if you add more than that increment your valuation starts losing more and more accuracy (because for each variable the valuation actually depends on that variable, savvy?). I mean, while people dish out the idea that 14RAP = 1DPS the reality is that the RAP-DPS graph is a curve when taking fluctuating crit values into consideration (14RAP at x crit is really zDPS, but at y crit it is aDPS). With the curves you could get a feel for relative slopes and figure out at any give point when to shift your focus for itemization.
Does something like that sound useful to you guys? I kinda think so, since it would be gear-independent and allow for a better sense of what kind of upgrades are more beneficial (instead of what specific upgrade is best buy guess and checking it in one of the inclusive spreadsheets).
Update: I think I settled on the best visualization for it. A single 3d graph with DPS, Crit %(rating?), and RAP as the axises. It'd take the same inputs as my spreadsheet plus a shot cycle/duration and pop out the proper graph. For spice the agi curves for each of the buff-states could be drawn on as well. Maybe allow for the suspension of one axis (dropping it down to DPS and Crit for instance) and the isolation of the agi curves (so the axises would be DPS and agi). Hmm...yes, I think those would be interesting. Could a spreadsheet do that? Or would it need to be a 3d graphing tool?
|
You're essentially asking for plane curves that lie in a plane that crosses the AP/CRIT/HIT axes at certain values. For each shot, it should be possible to determine the formula dictating damage (this is done already for pretty much all shots as it stands). If nothing else, having the partial derivatives of each of the major players (ap/crit/hit) in the equation wouldn't be too difficult. Honestly, though, given how dependent hunter dps is on the sequence of abilities rather than maximizing the individual ability, I'm not sure how much practical value this would have.
|
|
|
|
|
06/07/07, 8:12 PM
|
#274
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
Originally Posted by The Iron Colonel
You're essentially asking for plane curves that lie in a plane that crosses the AP/CRIT/HIT axes at certain values. For each shot, it should be possible to determine the formula dictating damage (this is done already for pretty much all shots as it stands). If nothing else, having the partial derivatives of each of the major players (ap/crit/hit) in the equation wouldn't be too difficult. Honestly, though, given how dependent hunter dps is on the sequence of abilities rather than maximizing the individual ability, I'm not sure how much practical value this would have.
|
That's why I was saying to have the program take a shot cylce as input (I'm thinking # of each shot and time for cycle would work). So something like:
Variable Equations X = Change in Crit %, Y = Change in RAP
Variable Auto Shot Damage ((Crit % + X) * (Crit Multiplier) + (Hit % - (Crit % +X))) * (Average Weapon Damage + Scope + Ammo Damage + (((RAP + Y) / 14) * Weapon Speed))
Variable Steady Shot Damage ((Crit % + X) * (Crit Multiplier) + (Hit % - (Crit % + X))) * ((.2 * (RAP + Y)) + 150 + (Weapon DPS * 2.8))
Variable Arcane Shot Damage ((Crit % + X) * (Crit Multiplier) + (Hit % - (Crit % + X))) * ((.15 * (RAP + Y)) + 273)
Variable Multi Shot Damage ((Crit % + X) * (Crit Multiplier) + (Hit % - (Crit % + X))) * (Average Weapon Damage + Scope + Ammo Damage + 205 + (((RAP + Y) / 14) * 2.8))
DPS Equation ((# Auto Shots) * (Variable Auto Shot Damage) + (# Steady Shots) * (Variable Steady Shot Damage) + (# Arcane Shots) * (Variable Arcane Shot Damage) + (# Multi Shots) * (Variable Multi Shot Damage)) / (Seconds To Complete Cycle)
Graph the final DPS equation with X and Y as the variables and all other values being user input.
On a separate note, do damage multipliers just get multiplied on at the end? So the variable equation for auto shot including damage multipliers would be the following?
[((Crit % + X) * (Crit Multiplier) + (Hit % - (Crit % +X))) * (Average Weapon Damage + Scope + Ammo Damage + (((RAP + Y) / 14) * Weapon Speed))] * Damage Multiplier
edit: Oh, and I'm not asking for someone else to do it, I'm asking if anyone would be interested in that kind of tool, I'd be more than happy to make it since I'm kinda hankering for a side project to work on in my free time :P
|
|
|
|
|
06/07/07, 8:38 PM
|
#275
|
|
Don Flamenco
|
Ah ok, I misread you the first time. If it's done with respect to rotation input, then that might be a handy tool to have. I don't really have time or ability to aid in authoring any side projects currently (started managing a new project at work...kind of eats up time), but I'd be glad to help with the math in any way possible.
|
|
|
|
|
06/07/07, 8:48 PM
|
#276
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
If you could spot check the equations I laid out on the right here that would be awesome. Also, if you could confirm the way damage modifiers are included that would be awesome.
And by you I mean The Iron Colonel or Cheeky or anyone else that has dealt with these equations before 
|
|
|
|
|
06/07/07, 11:00 PM
|
#277
|
|
Don Flamenco
|
I don't have time tonight to thoroughly cross check the formulas for specific shots, but one thing I noticed is that you've calculated both autos and specials as having two tables for results (i.e. hits aren't crits, crits aren't hits). I'm fairly certain autos use 1 table (all crits are hits, all hits are not crits) and specials use 2 tables (crits aren't hits, hits aren't crits). It's kind of late, though, and I can't find a reliable source for this information. Can someone confirm or disconfirm this?
|
|
|
|
|
06/07/07, 11:30 PM
|
#278
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
Did I?
What I'm doing is taking the base damage of a shot, and multiplying it by
[(Crit%*CritMod) + (hit%-crit%)]
So that distributes to (with X representing base damage)
(Crit%*CritMod*X) + (hit%-crit%)*X
or with 100% chance to hit
(Crit%*CritMod*X) + (1-crit%)*X
Is that not correct? I wana say that's how Cheeky does it in his spreadsheet (and I think he verified that part for me), but I could be wrong too :P
|
|
|
|
|
06/08/07, 1:20 AM
|
#279
|
|
Von Kaiser
|
-nt
Last edited by Xeno : 06/08/07 at 1:26 AM.
|
|
|
|
|
06/08/07, 1:27 AM
|
#280
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
Originally Posted by Xeno
Groggan in your spreadsheets default values I noticed you entered 2.33 as your crit modifier. I think 2.37 would be more accurate, as the Relentless Earthstorm Diamond affects the total crit, not the crit bonus like mortal shots.
|
I don't have that diamond, was trying to account for slaying talents, but I see in the earthstorm diamond thread that slaying works differently too, so I'll probably adjust it to 2.44 (2.3 *1.03 *1.03) and drop the damage modifier down to 1.03 (I'm looking at damage with FI up also).
edit: republished with those changes.
edit: Can anyone take a look at the variable equations and dps equations and confirm they are correct? Particularly the dps equation. I wana get it right before I take the time to add in the shot cycle inputs etc.
|
|
|
|
|
06/08/07, 3:06 AM
|
#281
|
|
Don Lactose
Tauren Hunter
Talnivarr (EU)
|
I don't really have time to check this, going to work in a sec.
However, there are some formulae here you might want to take a look at:
http://hunterguide.wikispaces.com/Theorycraft
|
Look, Lactose, we'd rather you didn't eradicate the whole human race.
- Sam & Max
|
|
|
06/08/07, 9:26 AM
|
#282
|
|
Don Flamenco
|
Originally Posted by Groggan
Did I?
What I'm doing is taking the base damage of a shot, and multiplying it by
[(Crit%*CritMod) + (hit%-crit%)]
So that distributes to (with X representing base damage)
(Crit%*CritMod*X) + (hit%-crit%)*X
or with 100% chance to hit
(Crit%*CritMod*X) + (1-crit%)*X
Is that not correct? I wana say that's how Cheeky does it in his spreadsheet (and I think he verified that part for me), but I could be wrong too :P
|
That distributes for a one roll system (crits are not hits, hits are not crits). IF it uses a two roll system (crits are hits, not all hits are crits), critical strikes are the product of hit and crit. This would look like:
[(hit%-hit%*crit%)+hit%*(crit%*CritMod)]*DamageFromShot
So the sum of hits and crits is equal to the hit%, which may be less than 100% of all attacks (it doesn't account for misses, but with 100% to hit it is the same as your last formula, which we would expect). Anyway, you could simplify that equation but I think that form is most demonstrative of my point. This is why I asked; again, I haven't confirmed it's a two roll special / one roll auto system yet, but that's the mathematical line of reasoning with which I approached it.
|
|
|
|
|
06/08/07, 11:52 AM
|
#283
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
I'd never heard anyone mention that any of our specials are on the two-roll system. I though they were all treated the same as auto shot (which is treated the same as melee attacks, including specials, which are one-roll).
Anyone know for sure?
|
|
|
|
|
06/08/07, 12:37 PM
|
#284
|
|
Don Flamenco
|
There's some good info in [dps Warrior] Hit is hugely overrated if you want to read more. It appears that the consensus is that it is a 1 roll system, but I don't want to put words in mouths. Essentially, this is what the info in that thread says:
* There is 1 roll to determine if something is a hit, miss, crit, glance, etc.
* 1% crit has no effect on the miss %, but subtracts 1% from your chance to hit
* 1% hit has no effect on the crit %, but subtracts 1% from your chance to miss
Make sense? I think I have that correct. Assuming it is, the formula should be
(ChanceHit+Crit%*CritModifier)*ShotDamage = TotalDamageFromShot
where
ChanceHit = 1 - (ChanceMiss - %Hit) - Crit%
ShotDamage = whatever the shot damage formula is for the given shot
So your Hit% converts miss chances to hits, and crit% converts hit chances to crits. I believe this is correct, but again, math mistakes may have been made.
|
|
|
|
|
06/08/07, 1:22 PM
|
#285
|
|
Piston Honda
Blood Elf Death Knight
Feathermoon
|
I think you are complicating it there. For a 1-roll system you can just do what I'm doing and use:
(Crit%*CritMod)*Damge + (Hit%-Crit%)*Damage
What this does is for your %crit multiplies the damage by the crit mod, then it takes and subtracts out your crits from your total hits and multiplies that % by shot damage. Or if you reverse the order maybe it would make more sense:
(Hit%-Crit%)*Damage + (Crit%*CritMod)*Damge
So it's how many hits will I get that are not crits, and then how many crits will I get. If you add up (Hit%-Crit%) and Crit% you of course get Hit%, and if you subtract that from 100% you get your miss%, which is in line with a 1-roll system.
|
|
|
|
|
|