Elemental Wotlk Talents discussion - Page 47

dedmonwakeen · 10/04/08, 6:41 PM

Originally Posted by Roywyn

Here is a link with a full Output for Simcraft at the latest patch:
SampleOutput - simulationcraft - Google Code
Bear in mind that mage DPS is likely 12% too high across the board in the sim due to Torment the Weak not properly triggering on bosses (and likely not being intended to trigger either).

The Shaman is only using LB/CL/LvB. I can talk to Deadmon to modify the castsequence, but would need some more info:
Coefficient confirm on LB/CL/LvB. Coefficients for FS DD and FS DoT.
Cooldowns between spells, and which spells share cooldowns (Shocks and LvB do, no)?
Crit damage of Lava Burst with (or without) CSD fully talented.
A spec (mmo-champion link).
Any glyphs that matter.

A cast sequence/priority list. Should be:
FS - then cast CL (if off cooldown) and LB - LvB when you cut off one tick.

With that info, we should be able to sim Elemental Shaman better.

I'm afraid the Elemental action list was vicitimized by swapping the JoW cooldown between player-based and target-based multiple times. Without JoW the Elemental Shaman was having serious mana issues. (of course, i sim a full burn with no rest periods, so.....)

Anyway, the sample output that Roywyn linked has been updated. (as well as the raid_80.txt config file that can be found at the same place). However..... if at the moment if your were to grab the windows download it would contain a stale version of the config.

The results with Flame Shock added:

Clicking through to Sample Output will show a complete breakdown for each class/spec combo.

I appreciate the peer review for our Shaman modeling. It hasn't been scrutinized to the same degree as the other supported classes.

Juice · 10/04/08, 8:57 PM

Dear My favorite thread:

We're getting a flurry of reports from here as people start to disagree with each other regarding the math and theory surrounding elemental shaman. Some of the reports are borderline, some are absurd. So here's an over-arching comment: Keep it polite.

We are on the same side here, we can discuss the issue without trolling, getting our feelings hurt, or calling people's contributions and intelligence into question. Each of the people involved in this is a valuable contributor here, so I'd like to see you self police this a bit more.

Binkenstein · 10/04/08, 9:10 PM

dedmon, question re the elemental stats.

#
# Elemental Shaman
#
# (1) The "trigger=3500" option to mana_potion represents the mana gained and the amount below max required to trigger
# (2) The "flame_shock=1,max_ticks_consumed=1" option to lava_burst will prevent it from being called until there is only one tick remaining on Flame Shock
#
shaman=Shaman_Elemental
level=80
talents=http://talent.mmo-champion.com/?shaman=353005150021535102501350000000000000000000000000003050500300000000000000000000
actions=flask,type=blinding_light/flametongue_weapon,weapon=main/mana_potion,trigger=3500/mana_spring_totem/totem_of_wrath/wrath_of_air_totem/elemental_mastery/flame_shock/lava_burst,flame_shock=1,max_ticks_consumed=1/bloodlust,target_pct=35/chain_lightning/lightning_bolt
gear_stamina=708
gear_intellect=777
gear_spirit=411
gear_spell_power=2200
gear_crit_rating=461
gear_hit_rating=250
gear_haste_rating=492
glyph_flametongue_weapon=1
glyph_totem_of_wrath=1
glyph_lightning_bolt=1
# Throw in a pair of Paladins.......  Eventually these static parms will be replaced with real actors.
blessing_of_kings=1
blessing_of_wisdom=1
sanctified_retribution=1
swift_retribution=1

Are the stat figures there including or excluding buffs, I assume that class specific buffs/debuffs are included but only those specified from other classes are?

dedmonwakeen · 10/05/08, 1:05 AM

Originally Posted by Binkenstein

dedmon, question re the elemental stats.

Are the stat figures there including or excluding buffs, I assume that class specific buffs/debuffs are included but only those specified from other classes are?

The the gear_xxx stats are simply that...... just gear. Provided I have a class implemented in the sim, the only way to get a buff/debuff from that class is to actually include it in the raid. If you want MotW, include a Druid. If you want ImprovedScorch/WintersChill, bring a Mage.

The exceptions to this are buffs from classes I have not yet implemented, such as Paladins...... which is why you will see swift_retribution=1 sprinkled all through the config file. Paladins are next on the list to be added to the simulator...... just because I can't stand all the unnecessary noise in the config file.

The actual gear values were from someone who did a first pass at a Naxx10 kit. I believe Roywyn has done considerable work in this area to refine this data, so an overhaul of the config file is probably in order. (Not only that, but the Mage and Warlock forums have contributed very helpful recommendations on talent/action tweaks that should be included.)

Binkenstein · 10/05/08, 2:26 AM

Hmmm, if that's the case your elemental model is off, although I'm using top of the line gear.
1006 stam, 902 int, 2263 spellpower, 416 crit, 277 hit, 81 mp5, 475 haste (enchants haven't been updated yet)
4975 dps, LvB/FS/CL/LBx3, 6.4 minutes
4814 dps, LvB/FS/LBx4, 132 minutes

This is using same glyphs, one heroism, a 40ms delay, and full raid buffs.
http://talent.mmo-champion.com/?sham...00000000000000
Doesn't include the set bonuses either, and only +12 spellpower ring enchants (no other tradeskills in effect)

[e]link to spreadsheet

dedmonwakeen · 10/05/08, 3:22 AM

Originally Posted by Binkenstein

Hmmm, if that's the case your elemental model is off, although I'm using top of the line gear.
1006 stam, 902 int, 2263 spellpower, 416 crit, 277 hit, 81 mp5, 475 haste (enchants haven't been updated yet)
4975 dps, LvB/FS/CL/LBx3, 6.4 minutes
4814 dps, LvB/FS/LBx4, 132 minutes

This is using same glyphs, one heroism, a 40ms delay, and full raid buffs.
http://talent.mmo-champion.com/?sham...00000000000000
Doesn't include the set bonuses either, and only +12 spellpower ring enchants (no other tradeskills in effect)

[e]link to spreadsheet

Thanks for the analysis. I'll dig in. Of the top of my head, the biggest difference is that I use an average lag of 200ms. I say "average" because in reality it is a random number between 150ms and 250ms. (The range is fixed, the "average" value can be modified.)

The next biggest thing would probably be time spent refreshing totems.

Dhiva · 10/05/08, 3:45 AM

Thanks for the analysis. I'll dig in. Of the top of my head, the biggest difference is that I use an average lag of 200ms. I say "average" because in reality it is a random number between 150ms and 250ms. (The range is fixed, the "average" value can be modified.)

The next biggest thing would probably be time spent refreshing totems.

Thanks for looking it into it, I'm currently trying out simu craft and I was kinda worried

Usually, I'm somewhere like, 250-500 ms ping from the server, however, using LagInfo, my average time between casts stays aroung 25-50ms. On the other hand, chain lighning are around 100ms between casts.

So I was wondering how accurate 0.2 sec between cast is ?

Binkenstein · 10/05/08, 4:19 AM

A delay of 200ms is a rather large delay given the "spell queue" system. Using lag info I can get sub 80 usually.
Totems should be ~4-6 seconds every 5 minutes now, so it's not much of a difference.

I think the big thing is getting the flame shock/lava burst interaction down, as there's no way in hell you should be seeing chain lightning give roughly the same damage (or even exactly the same).

dedmonwakeen · 10/05/08, 8:50 AM

Originally Posted by Binkenstein

A delay of 200ms is a rather large delay given the "spell queue" system. Using lag info I can get sub 80 usually.
Totems should be ~4-6 seconds every 5 minutes now, so it's not much of a difference.

I think the big thing is getting the flame shock/lava burst interaction down, as there's no way in hell you should be seeing chain lightning give roughly the same damage (or even exactly the same).

Is wrath of Air still at 2min? Did that one get changed, too?

Judging by the average counts below, it looks like the priority list is missing 1LvB per 10FS. I only allow the LvB to trigger if by the end of the cast it would only eat 1 tick of FS. But....... this is not the same thing as a fixed rotation that you have specified. If casting a LB right NOW means that I won't be able to get the LvB in right AFTER is it better to LvB right now even if it means eating 2 ticks of FS?

Also, in my gear/raid setup, I'm seeing CL give a dmg-per-cast of 5242 and LB is at 4777. Is this still in your "roughly the same damage" definition? If so, what kind of gap would you expect? (I'll be able to try out your spreadsheet later...... but it will be a while.)

Player=Shaman_Elemental  DPS=3479.9  DPR=13.0  RPS=266.8/209.0  (mana)
  Core Stats:  strength=138  agility=76  stamina=927  intellect=990  spirit=606  health=12278  mana=19694
  Spell Stats:  power=2200  hit=9.5%  crit=18.2%  penetration=0  haste=15.0%  mp5=91
  Attack Stats:  power=333  hit=7.6%  crit=12.6%  expertise=0.0  penetration=0  haste=15.0%
  Actions:
    chain_lightning       Count= 47.7| 6.1sec  DPE=5242|25%  DPET=3835  DPR= 7.3  Miss=1.5%  Hit=3554  CritHit=7107|7107|49.0%
    flame_shock           Count= 30.3| 9.6sec  DPE=3582|11%  DPET=2621  DPR= 8.7  Miss=1.4%  Hit=1246  CritHit=2492|2492|43.8%  Tick=473
    lava_burst            Count= 27.1|10.7sec  DPE=7377|20%  DPET=5404  DPR=25.4  Miss=1.5%  Hit=3436  CritHit=7491|7491|98.5%
    lightning_bolt        Count= 80.2| 3.6sec  DPE=4777|38%  DPET=2771  DPR=18.5  Miss=1.5%  Hit=3238  CritHit=6477|6477|49.0%
    lightning_overload    Count= 31.3| 9.3sec  DPE=2202| 7%  DPET= inf  DPR= inf  Miss=1.3%  Hit=1498  CritHit=2993|3554|48.5%

Juice · 10/05/08, 12:33 PM

Wrath of air is now at 5 minutes. All previous 2m totems are now 5m, basically.

Ocyr · 10/05/08, 12:48 PM

Originally Posted by dedmonwakeen

If casting a LB right NOW means that I won't be able to get the LvB in right AFTER is it better to LvB right now even if it means eating 2 ticks of FS?

edit: after I factored in the haste, yes I can see how Lava Burst might chop off a tick of Flame Shock. Is there any reason not to cast an extra Lightning Bolt in the rotation, in those situations?

Binkenstein · 10/05/08, 2:56 PM

Originally Posted by dedmonwakeen

Is wrath of Air still at 2min? Did that one get changed, too?

Judging by the average counts below, it looks like the priority list is missing 1LvB per 10FS. I only allow the LvB to trigger if by the end of the cast it would only eat 1 tick of FS. But....... this is not the same thing as a fixed rotation that you have specified. If casting a LB right NOW means that I won't be able to get the LvB in right AFTER is it better to LvB right now even if it means eating 2 ticks of FS?

Also, in my gear/raid setup, I'm seeing CL give a dmg-per-cast of 5242 and LB is at 4777. Is this still in your "roughly the same damage" definition? If so, what kind of gap would you expect? (I'll be able to try out your spreadsheet later...... but it will be a while.)

Player=Shaman_Elemental  DPS=3479.9  DPR=13.0  RPS=266.8/209.0  (mana)
  Core Stats:  strength=138  agility=76  stamina=927  intellect=990  spirit=606  health=12278  mana=19694
  Spell Stats:  power=2200  hit=9.5%  crit=18.2%  penetration=0  haste=15.0%  mp5=91
  Attack Stats:  power=333  hit=7.6%  crit=12.6%  expertise=0.0  penetration=0  haste=15.0%
  Actions:
    chain_lightning       Count= 47.7| 6.1sec  DPE=5242|25%  DPET=3835  DPR= 7.3  Miss=1.5%  Hit=3554  CritHit=7107|7107|49.0%
    flame_shock           Count= 30.3| 9.6sec  DPE=3582|11%  DPET=2621  DPR= 8.7  Miss=1.4%  Hit=1246  CritHit=2492|2492|43.8%  Tick=473
    lava_burst            Count= 27.1|10.7sec  DPE=7377|20%  DPET=5404  DPR=25.4  Miss=1.5%  Hit=3436  CritHit=7491|7491|98.5%
    lightning_bolt        Count= 80.2| 3.6sec  DPE=4777|38%  DPET=2771  DPR=18.5  Miss=1.5%  Hit=3238  CritHit=6477|6477|49.0%
    lightning_overload    Count= 31.3| 9.3sec  DPE=2202| 7%  DPET= inf  DPR= inf  Miss=1.3%  Hit=1498  CritHit=2993|3554|48.5%

The ideal rotation should be LvB/FS/LB/LB/LB/LB, replacing the first LB with CL if you wish (may want to run two sims, one with and one without CL).
Add another 1.5% hit in there, because everyone will be hit capped no matter what (40 hit from food ~ 1.5%)
Also, the spec to use should be http://talent.mmo-champion.com/?sham...00000000000000 mostly because Enh totems is useless (doesn't affect ToW), Reverb is pointless (only cast FS after LvB, which has a longer cooldown), and Thunderstorm is ~90-100 mp5 at those gear levels.
[e]May want to re-check the buffs you're using as well, Flametongue should be 274 spellpower talented.

Dhiva · 10/05/08, 4:24 PM

Originally Posted by Binkenstein

The ideal rotation should be LvB/FS/LB/LB/LB/LB

Though we probably wouldnt start with LvB when the fight is first engaged, right ?
So that might be an exception to add in the code.

Binkenstein · 10/05/08, 5:40 PM

Originally Posted by Dhiva

Though we probably wouldnt start with LvB when the fight is first engaged, right ?
So that might be an exception to add in the code.

Well, FS/LB/LB/LB/LB/LvB if you want to get picky.

# of LB = LB Cast time / (8 - GCD), rounding up to the nearest whole number. Subtract two GCD times if you are using CL as well.

[e]Also, ignoring cast time LB should always be higher damage than CL except at low gear levels.

dedmonwakeen · 10/06/08, 1:59 AM

Binke,

I took a look at your spreadsheet and these are some of the differences I found:

(0) Lag (I just set mine to 40ms)
(1) I had the "old" coef for CL.
(2) Gear level: Your gear level was a bit higher (except for Int which was much higher)
(3) You had chaotic skyfire diamond while i did not
(2) Focus Magic: I was only allowing the Mages this buff
(3) Tier6 4-Pc Bonus: I did not have this on..... It appears you do, but I cannot be sure.
(4) Very minute differences in coef for FS
(5) Minor difference in haste merging. I do 1.0/( Haste1 * Haste2 * Haste3). You seem to add them and then do 1.0/(1.0 + HasteSum).
(6) Totems: I added a second Shaman to the raid that did not have to worry about putting down totems
(7) I had the old behavior of Ancestral Knowledge
(8) 20D on the DEP tab: I can't figure out how your buffed crit gets so high
(9) 32E on the DEP tab: Are you sure this is supposed to be 3? Shouldn't it be (5/6)=0.83 to represent 1 tick eaten by LvB?

Regarding #8
1209 Int => 7.3% (1209/166)
Thundering Strikes => 5%
Moonkin => 5%
Focus Magic => 3%
Improved Scorch => 10%
Totem of Wrath => 3%
FT Glyph => 2%

This only totals 35.3%....... You have 42.45%. What am I missing?

Due to uncertainties surrounding #8 and #9 I cannot get my simulator to match your spreadsheet.

EDIT: SimulationCraft version r680 available for download, including CL fix.

Ocyr · 10/06/08, 2:13 AM

Originally Posted by dedmonwakeen

(2) Gear level: Your gear level was a bit higher (except for Int which was much higher)
(4) Very minute differences in coef for FS
(6) Totems: I added a second Shaman to the raid that did not have to worry about putting down totems.

Regarding:

2) Are you simulating each class in their own gear, or in equal gear (they are not the same)?

4) binke's coef for flame shock was incorrect, I believe; the correct coef for flame shock has been provided by Tufy and others and was just slightly different from the coefs he listed (the coef listed in wowiki is wrong, there was some discussion about this a few pages back)

6) Why? It is not realistic to expect a raid to bring 2 Elemental Shaman. I think any realistic simulation of Elemental DPS will include totem dropping / GCD engagement.

dedmonwakeen · 10/06/08, 2:34 AM

Originally Posted by Ocyr

Regarding:

2) Are you simulating each class in their own gear, or in equal gear (they are not the same)?

4) binke's coef for flame shock was incorrect, I believe; the correct coef for flame shock has been provided by Tufy and others and was just slightly different from the coefs he listed (the coef listed in wowiki is wrong, there was some discussion about this a few pages back)

6) Why? It is not realistic to expect a raid to bring 2 Elemental Shaman. I think any realistic simulation of Elemental DPS will include totem dropping / GCD engagement.

#2 If I post a chart ranking dps it will always have casters using the same gear unless I have a really, really, really good reason to open myself for lots of argument

#4 Thank you! Could we get the new coeff listed in the OP?

#6 I was more worried about the vast difference between his spreadsheet and my sim. My goal was not to match wow-reality but to remove as many obvious differences between our modeling in the hopes of tracking down any bugs. I found several in the sim, in particular the coeff to CL and my handling of Ancestral Knowledge. (Both came about because i was TRYING to model both BC and WotLK at the same time..... but my code was turning into spaghetti.)

We are still pretty far off which is why I was looking for confirmation on #8 and #9. If #9 is behaving as it appears (remember i'm no expert on xls!), then Binke is vastly overestimating the contribution of the FS DoT. The discrepancy on #8 could be due the spreadsheet using the old level 70 Intellect-to-Spell-Crit ratio, but I'm not sure.....

sjogren · 10/06/08, 3:50 AM

Originally Posted by dedmonwakeen

#2 If I post a chart ranking dps it will always have casters using the same gear unless I have a really, really, really good reason to open myself for lots of argument

I understand what you're saying, but eg spirit is entirely useless for shamans. I suppose the shaman itemization would get mp5 instead which doesn't contribute to dps either, so maybe it works out decently anyway. It would be nice to compare the dps with every class in their best-in-slot sets, but that's a lot of extra work for what's probably a minor increase in precision.

Ocyr · 10/06/08, 4:25 AM

Originally Posted by dedmonwakeen

#4 Thank you! Could we get the new coeff listed in the OP? If #9 is behaving as it appears (remember i'm no expert on xls!), then Binke is vastly overestimating the contribution of the FS DoT.

I listed it in my OP; http://elitistjerks.com/922437-post1149.html and those numbers I took from an earlier discussion that Tufy had back on page 35 or so. binke's numbers are less than 1% different for the initial damage, and 3% or so different for the damage over time component (20 vs 21%, and 42 vs 39%), from the numbers Tufy stated. So not much of a difference.

Hainar · 10/06/08, 6:20 AM

Originally Posted by Binkenstein

Also, the spec to use should be http://talent.mmo-champion.com/?sham...00000000000000 mostly because Enh totems is useless (doesn't affect ToW), Reverb is pointless (only cast FS after LvB, which has a longer cooldown), and Thunderstorm is ~90-100 mp5 at those gear levels.

Why that spec given that thunderstorm is too stituational and too short a range for PvE right now - it works against the standing at max range notion. Surely it would be better to have 5/5 US as in most fights TS is not likely to be used at all.

tukez · 10/06/08, 8:00 AM

Originally Posted by dedmonwakeen

(5) Minor difference in haste merging. I do 1.0/( Haste1 * Haste2 * Haste3). You seem to add them and then do 1.0/(1.0 + HasteSum).

AFAIK, haste is calculated like this: speed / ( haste_ratings_sum * haste2 * haste3 ). So you sum all the haste ratings, not static % haste. For example with 5% haste from rating and 30% from bloodlust: speed / ( 1.05 * 1.30).

I'm the author of EnhSim and it seems like I could easily add few talents to make it useful for Ele shamans too. It already has spell casting with duration, because of the Maelstrom Weapon. Maybe I'll wait until Blizzard have fixed the talents though, to prevent extra work.

ZulazeeluIcecrown · 10/06/08, 8:12 AM

Originally Posted by Hainar

Why that spec given that thunderstorm is too stituational and too short a range for PvE right now - it works against the standing at max range notion. Surely it would be better to have 5/5 US as in most fights TS is not likely to be used at all.

Did you notice that Thunderstorm restores 5% of your mana whether it hits anything or not? Or am I wrong in thinking getting 5% mana back is worth a GCD?

dedmonwakeen · 10/06/08, 8:55 AM

Originally Posted by tukez

AFAIK, haste is calculated like this: speed / ( haste_ratings_sum * haste2 * haste3 ). So you sum all the haste ratings, not static % haste. For example with 5% haste from rating and 30% from bloodlust: speed / ( 1.05 * 1.30).

That is indeed how I have it coded it up. I should have written: 1.0 / ( SourceOfHaste1 * SourceOfHaste2 * SourceOfHaste3 )

where Haste Rating is one source of Haste.

I'm the author of EnhSim and it seems like I could easily add few talents to make it useful for Ele shamans too. It already has spell casting with duration, because of the Maelstrom Weapon. Maybe I'll wait until Blizzard have fixed the talents though, to prevent extra work.

Cool! Enhsim is a great tool and very, very, very fast: Perfect for EP calculation.

I've had to sacrifice too much to cover general class support in a convenient manner. Five down..... Five to go.....

Binkenstein · 10/06/08, 3:45 PM

Originally Posted by dedmonwakeen

Binke,

I took a look at your spreadsheet and these are some of the differences I found:

Lag (I just set mine to 40ms)
I had the "old" coef for CL.
Gear level: Your gear level was a bit higher (except for Int which was much higher)
You had chaotic skyfire diamond while i did not
Focus Magic: I was only allowing the Mages this buff
Tier6 4-Pc Bonus: I did not have this on..... It appears you do, but I cannot be sure.
Very minute differences in coef for FS
Minor difference in haste merging. I do 1.0/( Haste1 * Haste2 * Haste3). You seem to add them and then do 1.0/(1.0 + HasteSum).
Totems: I added a second Shaman to the raid that did not have to worry about putting down totems
I had the old behavior of Ancestral Knowledge
20D on the DEP tab: I can't figure out how your buffed crit gets so high
32E on the DEP tab: Are you sure this is supposed to be 3? Shouldn't it be (5/6)=0.83 to represent 1 tick eaten by LvB?

Regarding #8
1209 Int => 7.3% (1209/166)
Thundering Strikes => 5%
Moonkin => 5%
Focus Magic => 3%
Improved Scorch => 10%
Totem of Wrath => 3%
FT Glyph => 2%

This only totals 35.3%....... You have 42.45%. What am I missing?

Due to uncertainties surrounding #8 and #9 I cannot get my simulator to match your spreadsheet.

EDIT: SimulationCraft version r680 available for download, including CL fix.

3) Gear level wouldn't account for the differences.
4) Best meta, but it's easier to model things without tbh.
5) With the high crit from talents, Elementals will be the "best" target for that buff.
6) Support is there, but it should have been switched off (Check LB % Gain on DEP is 1.05)
7) FS co-eff, see bellow
8) Haste ratings add up. Heroism is the only % gain accounted for, and it goes Cast/1.3/(1+haste)
9) Best to rely on one elemental shaman, dropping 2 totems (ToW, Mana), as the resto will cover WoA for the raid.
11) I counted CoT & TS twice >.<
12) There are four ticks (five with the glyph). This is adjusted based on how many ticks you "clip" on the main page.

Originally Posted by Ocyr

4) binke's coef for flame shock was incorrect, I believe; the correct coef for flame shock has been provided by Tufy and others and was just slightly different from the coefs he listed (the coef listed in wowiki is wrong, there was some discussion about this a few pages back)

I'd be interested in seeing the calculations behind that, because I'm using the method listed here. So either I'm wrong, and that page is wrong (which means there's been a change in how DD/DoT spells are calculated), or Tufy got his calculations wrong.

Going by that, it's 1.5/3.5 * (500/(500+556)) which comes out at 0.429 * 0.474 which comes out at 0.203 for the direct damage portion, and 12/15 * (556/(500+556)) -> 0.8 * 0.527 = 0.421

Re: Thunderstorm. Based on current gear it gives me ~76 mp5. At 80 it will give me much more. It's like a lifetap/evocate hybrid, and if you run out of mana and don't have it, what are you going to do?

Ocyr · 10/06/08, 4:20 PM

Originally Posted by Binkenstein

I'd be interested in seeing the calculations behind that, because I'm using the method listed here. So either I'm wrong, and that page is wrong (which means there's been a change in how DD/DoT spells are calculated), or Tufy got his calculations wrong.

Going by that, it's 1.5/3.5 * (500/(500+556)) which comes out at 0.429 * 0.474 which comes out at 0.203 for the direct damage portion, and 12/15 * (556/(500+556)) -> 0.8 * 0.527 = 0.421

Daidalos post on the subject: http://elitistjerks.com/f31/t27016-e...30/#post859738

Post between Tufy and Ofancow verifying 21/39% are the actual coefs: http://elitistjerks.com/f31/t27016-e...iscussion/p35/

Looks like both the wowiki and the EJ listed coefs are incorrect. Possibly due to unlisted changes between the PTR and Live?