Elitist Jerks
Register
Blogs
Forums


Go Back   Elitist Jerks » Class Mechanics » Warriors

Reply
 
LinkBack Thread Tools
Old 05/29/09, 2:32 PM   #31
hellord
Absolute Arms
 
hellord's Avatar
 
Orc Warrior
 
Hellscream (EU)
Ok, so we were thinking to get 1.1697% of damage converted in rage, we instead have 0.8273% damage as rage.
Also you can never exceed 3.3090% damage as rage (7.5/453.3)*2.
So for 1000 damage you can't get more than 33 rage regardless of the weapon speed.

The cap for a 140 hit is 4.6 rage So I have to guess regardless of the roundings it will never be 5.
Since a 140 hit with a 3.2s weapon is (140*7.5/453.3 + 3.5*3.2)/2 the result is 6.75 then lowered to 4.

ArP Whore

Offline
Reply With Quote
Old 05/29/09, 3:55 PM   #32
NobleHelium
Von Kaiser
 
Night Elf Warrior
 
Uther
Never mind, I'm an idiot sorry.

Offline
Reply With Quote
Old 05/29/09, 4:03 PM   #33
catch22atplay
Piston Honda
 
Tauren Warrior
 
Proudmoore
Well i wish this was figured out quite a while ago. How long have we been 80 now? I know the question had been brought up before at least on the WoW forums. And thankfully Landsoul got blizzard to notice his post. And thankfully Blizzard responded with a good answer. This info should have been provided by Blizzard a while ago. Although i understand they aren't obligated to do so.

I haven't mathed it yet. But a few questions come to mind. 1st they didn't tell us exactly how they arrived at 453.3 for the constant. I'm curious what formula they use now for this. We hit for 1k, 2k, 3k etc. With a cap to how much rage we can get per x damage. How much of a difference in rage is there vs the old formula. Bearing in mind the cap also. And since TG got nerfed how much rage does a fury warrior get now vs before the TG nerf using the same new formula for both. 83%? Very curious about that last one.

This new formula effects a lot of programs and websites. Quite an impact overall.

Last edited by catch22atplay : 05/29/09 at 4:08 PM.

Offline
Reply With Quote
Old 05/29/09, 8:44 PM   #34
hellord
Absolute Arms
 
hellord's Avatar
 
Orc Warrior
 
Hellscream (EU)
Just a few numbers.
How much influence does crit have on rage generation? How much did we THINK it was?
With the old formula crit was 2X/2 rage from damage and 7*speed/2 rage due to weapon.
For a 2k HIT with a boh this was
old: (2000*7.5/320.6+3.4*3.5)/2 = ~29 rage
new: (2000*7.5/453.3+3.4*3.5)/2 = ~22 rage

A 2k crit would instead generate
old: (2000*7.5/320.6+3.4*7)/2 = ~35 rage
new: (2000*7.5/453.3+3.4*7)/2 = ~28 rage

The static part of the formula hasn't changed at all. This practically means that if you could trade crit for ap/arp/exp/hit or any % dmg increase, so that your average damage stays the same, the crit "build" would produce more rage than we tought thus slightly increasing the value of crit.

This proportion decreases from 6% @1k damage to 1% @ 19k damage, and even at 100k damage grants a 0.21% rage increase compared to the old formula commonly used in spreadsheets.

It is very different than the supposed 3/8 vs 1/2 ratio that affected both factors and was theorically increasing the rage gain from damage vs the rage gain from weapon speed factor, thus decreasing the contribution of crit to rage.

I didn't put these numbers on the spreadsheet yet, but I guess this could be a sort of buff for crit for low geared furies, since at low levels the static part has a high impact.
On the other side Arms get 25% more rage from damage done that for a 2k hit is a 18%sh rage increase. The higher the damage the greater the increase.

Few other numbers.
@ 11369 damage w/o Endless rage a hit will produce 100 rage.
@ 10650 damage w/o Endless rage a crit will produce 100 rage.
@ 9095 damage with Endless rage a hit will produce 100 rage.
@ 8520 damage with Endless rage a crit will produce 100 rage.

The cap is only used to limit the rage income at very low damages limiting it to 3% of your damage.
The cap for a 3.4s weapon is around 240 for hit and 480 for crit. Over that damage value you should generate rage normally.

ArP Whore

Offline
Reply With Quote
Old 05/30/09, 4:11 AM   #35
BWarner
Piston Honda
 
BWarner's Avatar
 
Human Warrior
 
Uldum
Originally Posted by ZombieRitual View Post
<snip GC quote>

Seems our formula was just off. I do find it annoying however that the client doesn't always show our true rage.
Agreed. This is extremely annoying, considering our shouts cost 10 rage apiece, and our ability-based rage gains are in even multiples of 10. As a result, the last 10 rage generated is always wasted, in terms of shouting. (Speaking out of combat, of course.)

What SEEMS to be happening is thus: A Warrior uses a talented (1/2) Berserker rage, which should, for at least one full second, give the Warrior 10 rage, allowing the use of one shout, which would completely consume all available rage. However, due to the new nature of energy "prediction" by the client, the Warrior, upon using Berserker Rage, instantly loses the first point of rage from the client "predicting" the loss of rage due to natural rage decay.

Originally Posted by catch22atplay
Well i wish this was figured out quite a while ago. How long have we been 80 now? I know the question had been brought up before at least on the WoW forums. And thankfully Landsoul got blizzard to notice his post. And thankfully Blizzard responded with a good answer. This info should have been provided by Blizzard a while ago. Although i understand they aren't obligated to do so.

I haven't mathed it yet. But a few questions come to mind. 1st they didn't tell us exactly how they arrived at 453.3 for the constant. I'm curious what formula they use now for this. We hit for 1k, 2k, 3k etc. With a cap to how much rage we can get per x damage. How much of a difference in rage is there vs the old formula. Bearing in mind the cap also. And since TG got nerfed how much rage does a fury warrior get now vs before the TG nerf using the same new formula for both. 83%? Very curious about that last one.

This new formula effects a lot of programs and websites. Quite an impact overall.
Did they ever tell us why they chose the previous, level 70, constant? Does it matter? Not particularly. That's the value they chose for balance purposes. Not saying that it's right or wrong, but that's the value that was selected, by whatever method.

Also, note that the formula itself (algebraic functions) did not change. The only thing that changed was the value of the constant.

Last edited by BWarner : 05/30/09 at 4:22 AM.

Offline
Reply With Quote
Old 05/30/09, 12:39 PM   #36
catch22atplay
Piston Honda
 
Tauren Warrior
 
Proudmoore
Which doesn't help if we don't know the value of c. I don't need an answer as to why they chose the L70 constant. Nor do i need to know why 453.3 was chosen for L80. I'd just like to know what the math is. How is the number for c determined mathematically. For spreadsheet purposes for people that are leveling they need to know what that constant is at each level. So right now we have an incomplete answer of the whole formula. We know what it is for L60, L70 and L80 now. Anything in between those levels are pure assumptions. Although i believe 1-59 was tested. 61-69 may have been tested. But 71-80 was not tested to confirm and verify previous math assumptions. ok we know 80 now. So scratch that off my 71-80 list. I can easily fill those numbers in based on known c values at 3 given levels. But i'd then have to test and confirm it. And it still may be wrong although close.

Since the formula for c is unknown how are websites and spreadsheets going to incorporate c. This has a huge impact on like 20+ programs and websites. It effects the tanking ones as well. Everything now needs to be redone. At the minimum at least L80 can be fixed. So that's better than nothing as that's the most important number. But accuracy at any and all levels would be the perfect answer.

Last edited by catch22atplay : 05/30/09 at 1:20 PM.

Offline
Reply With Quote
Old 05/30/09, 1:28 PM   #37
catch22atplay
Piston Honda
 
Tauren Warrior
 
Proudmoore
Originally Posted by hellord View Post
JThe static part of the formula hasn't changed at all. This practically means that if you could trade crit for ap/arp/exp/hit or any % dmg increase, so that your average damage stays the same, the crit "build" would produce more rage than we tought thus slightly increasing the value of crit.

This proportion decreases from 6% @1k damage to 1% @ 19k damage, and even at 100k damage grants a 0.21% rage increase compared to the old formula commonly used in spreadsheets.

It is very different than the supposed 3/8 vs 1/2 ratio that affected both factors and was theorically increasing the rage gain from damage vs the rage gain from weapon speed factor, thus decreasing the contribution of crit to rage.
This is exactly the answer i wanted and suspected. This more closely matches in game results. I suspect a lot of items values and your choices will now change based on this. I suspect that the loss of crit with Ulduar gear is why people feel their rage is low and suspected a rage nerf when that wasn't the case. They got new gear but they lost a bit of rage. Or in some cases did not get nearly the increase in rage and dps as they thought they should have compared to previous gear. The corrected formula has quite an impact.

Last edited by catch22atplay : 05/30/09 at 1:47 PM.

Offline
Reply With Quote
Old 05/30/09, 5:38 PM   #38
Thymoleon
Glass Joe
 
Thymoleon's Avatar
 
Orc Warrior
 
Lightbringer (EU)
If we know the constant's values at L60,70,80, that's enough to calculate a linear regression, a quadratic and an exponential one. I can't remember the L60 c, but I suppose it's pretty easy to find, and calculating the regressios will take a minute. Then a lot of testing would be nedeed to state which is the better. An approximate formula with an error around 5% would be very satisfaing.

Offline
Reply With Quote
Old 05/30/09, 7:37 PM   #39
Machinator
Don Flamenco
 
Troll Warrior
 
Aggramar
Originally Posted by Thymoleon View Post
If we know the constant's values at L60,70,80, that's enough to calculate a linear regression, a quadratic and an exponential one. I can't remember the L60 c, but I suppose it's pretty easy to find, and calculating the regressios will take a minute. Then a lot of testing would be nedeed to state which is the better. An approximate formula with an error around 5% would be very satisfaing.
We had the formula for finding c derived already(its still on wowwiki), thats is where we had the 320.6 from in the first place. The problem is that it is not the same formula past 70, so no regression will be accurate. As far as I know the formula for c past lvl 70 has not been released by blizz and finding data points for it would be a huge pain.

"Information is ammunition."

Offline
Reply With Quote
Old 05/30/09, 8:16 PM   #40
landsoul
Myrmidon Champion
 
landsoul's Avatar
 
Worgen Warrior
 
Alterac Mountains
If you found the regression model it wouldn't even matter, because they would change it again in the next expansion and we would run into the same problem. At least now we know what to expect and how to test it to get the right c value.

>--Coaching Site--< Private coaching / Warrior Resource
>--Stream--< Tues, Wed, 7 Eastern.

Offline
Reply With Quote
Old 06/01/09, 6:00 AM   #41
ZorbaTHut
Von Kaiser
 
ZorbaTHut's Avatar
 
Troll Warrior
 
Mal'Ganis
Originally Posted by Thymoleon View Post
If we know the constant's values at L60,70,80, that's enough to calculate a linear regression, a quadratic and an exponential one. I can't remember the L60 c, but I suppose it's pretty easy to find, and calculating the regressios will take a minute. Then a lot of testing would be nedeed to state which is the better. An approximate formula with an error around 5% would be very satisfaing.
I don't really see why it should be assumed that any of those regressions will fit properly. I'd personally expect to see major slope discontinuities at 60 and 70. If we really wanted to get perfectly-accurate data, I think the best way would be to write a little UI mod with instructions to do the info gathering, put it out in the community, and ask people to go beat on training dummies for a bit with different-leveled warriors and submit their data.

Offline
Reply With Quote
Reply

Go Back   Elitist Jerks » Class Mechanics » Warriors

Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
2.3 UI changed Balroy The Dung Heap 0 11/13/07 5:27 PM
Rage generation and the Prot Warrior Splatter Class Mechanics 111 06/21/07 5:03 AM
Rage generation normalization in the Expansion Gyshall Public Discussion 536 10/06/06 2:21 PM