I could be wrong but my understanding is that in WoW we go from 0 to 60 in 0.0, there's no ramp up time, which makes the increased speed invaluable for getting out of the way.

Acceleration is instant, but it takes a few seconds for the difference to overcome the effects of lag/etc. and produce a meaningful visual difference.

In fact, tornhoof was the one who send me to this forum after i proved, that his calculations where not valid anymore

Doesn't these values make hit go far above the cap?

Tell me if there's errors in this line of reasoning
- Dual wield has a 24% base miss rate.
- You'll always have 1% miss rate, making hit useless after 23%
- 50 hit rating = 50 / 15.8 = 3.16% hit
- To cap hit rating from 22% hit you need 1% hit
- That means 2.15% hit was wasted, or that only 1/3.16 = 32% of the 50 HR had any effect

That's assuming an equal level target. Even if the target is 3 levels higher 50 hit rating would be enough to go beyond the cap. (Afaik 3 levels difference would only mean another 0.6% miss chance).

What am I missing?

Actually you didn't proof me wrong, since my Windfury calculations are based on PTR values and I stated that I don't know of any proper working closed form expression to model Windfury properly.

I also previously stated, that your model is flawed:
a.) you assume an even distribution of Windfury procs, which does not work simply due to the 3s cooldown, you need to find a distribution which covers the cooldown it. As soon as you move into a more discrete approach, you will enter LaPlace Probability. The correct way to model Windfury is probably to use a Markov-Chain, feel free to try this and come up with a closed form expression.

b.) You're using floor, there is no place for floor in a continuous model like you used, you're trying to fit your model to your simulation data, which simply does not work properly.


The only thing you came up with, which I don't use in my closed-form expressions is your preference and succinct usage of a factor to increase mainhand windfury damage.

While this is certainly true, I still think that without a proper WF modelling you can not assume that MH gets 56.9% of the Windfury procs.

The only thing you showed up to now, is that your system might hint that using a slightly faster MH might be preferable to a slower MH. As for now unfortunately this is not even possible to proof in the game due to the flurry mechanics.

Feel free to calculate an ideal model, feel free to calculate the ideal weapon speed ombination of MH and OH, as for now the ideal Speed is still for both as slow as possible.

You should compare your input data with the simualtion approaches of Pater and Disquette, my closed-form expression is simplified in some parts, since the goal of the closed-form expression was not to design an ideal model, but to create a model which is fast, to use it in a tool to optimize equipment.

Handuke -- the hit cap is 24% against even-level, 25.5% against L73, so we were under the cap, at 25.16%.

I have read this whole thread and of course the new summary thread too, but at the moment I am a bit baffled about my rather lackluster in-game results after switching to Enhancement two weeks ago.

I did some solo testing on the Blasted Lands mobs (albeit only 10 minutes stints with nothing else but WF on both weapons) and to my surprise my old combo of Decapitator (Mongoose) with Malchazeen (+7 Dmg) resulted in a DPS of 606, while my new combo of Decapitator (Mongoose) + Merciless Gladiator's Pummeler (no enchant at that time) only reached 580 DPS.

Later that night we killed Gruul and despite being raid buffed (including Kings and BEM flask) and having no positional or other problems during the fight I was only able to get 605 DPS.

My armory page:
http://armory.wow-europe.com/charact...Azshara&n=Zorc

Gruul WWS:
http://www.lossendil.com/wws/?report...d2cy63&s=0-326
(without the Crusader OH enchant and using Fetish of the Sand Reaver instead of Abacus of Violent Odds)

So my first question is: Why am I unable to reproduce the often quoted 'fact' that a fast OH is clearly worse than a slow OH ?

And the second is: Do I only need to increase STR/AP/Crit and the damage will come by itself or am I missing something about items or theory here ?

As I said I am skilled this way for no longer than two weeks and therefore some items I am wearing are rather subpar. My first goal was to reduce my miss rate, but I guess thats not worth much if my hits are missing the punch. I also thought wielding two of the best weapons at that level would lead to better results, even without having Kara epics on every slot. And frankly the difference between what I am wearing now and the heroic rewards or Kara drops does not seem that big. But I guess it all adds up, or does it ?

I don't think you're doing anything wrong. Other than slightly lower damage, you're doing about the same as I did when I came in 5th on Gruul recently.
http://www.lossendil.com/wws/?report...3275-3698&a=27

I assume that you just have very well-geared and skilled guildmates. I think they're just kicking ass, not that you're doing anything wrong.

Were you fully buffed with Might, BS, too?

Your DPS against the Blasted Lands mobs is a horrible benchmark. Your crit rate will generally be double what it is against level 70 mobs (probably resulting in more Flurry uptime for the Malch offhand, thus inflated DPS) and you will almost never miss even without much hit rating, and I highly suspect your WF proc rate is inflated against lower level mobs, though I've never confirmed it. Don't use that as a basis of comparison against a boss kill, it's apples and oranges. You're better off fighting some of the level 70 mobs in Netherstorm or Shadowmoon to compare DPS between weapon sets.

Honestly having Crusader on your offhand and the Abacus instead of the Fetish will probably make the difference of 20-25 DPS, especially if you are coordinating your cooldowns. Are you really that threat limited on Gruul that you need to use the Fetish? Looking at your WWS I have a tough time believing that's the case. Were you missing some buffs from previous attempts (i.e. BoMight)? I know there is a marked difference in my DPS depending on how many paladins we have show up. I think you're on the right track for gear and DPS (it's about what I do in comparable gear), don't let a single parse dissuade you. What you have will scale better as the rest of your gear catches up to your weapons.

10 minutes is a pretty small sample size. We rely so much on crits and WF procs to deal damage that our damage over a small period of time will vary by quite a bit. If you want to do a solid comparison, you're going to need an hour or more of test time with each weapon combination.

Unfortunately, the only mobs that will just let you beat on them for an hour are Blasted Lands mobs, which, as Nite_Moogle pointed out, might skew your results by increasing your ratio of white damage:windfury damage.

That's the reason there's so much debate over optimal weapon speed combinations and the like: we can't perform reliable DPS tests, so we're stuck with theorycraft, and it's hard to get clear results from theorycraft when no one is sure how flurry or windfury works.

I pretty much concur with whats posted above. The other thing to remember is that you were (rightfully so) using WF instead of GOA, had you used the latter it would have boosted your DPS closer to 680ish in my experience.

On the last Lurker fight (4 attempts, 1 kill), I had about 22% misses (combined) but 39% crit on average, but only 29% WF crits, which made me wonder a bit; I've heared of a crit cap, is this is at 50%? And secondly is the fact that the WF is showing a lower crit % due to a different hit-table?

Here's a Gruul WWS that I'm pretty happy with (915 dps with no gear outside Karazhan/Heroics): http://www.lossendil.com/wws/?report...=3743-4101&a=9 I had BS, LoTP, an Unstable Flask of the Beast, and a Survival Hunter in the raid. Ignore the fact that I cast Chain Lightning please, I messed up and hit the wrong button. >.<

Also, note I have very high +hit, and relatively low AP/Crit (26% hit, 22% crit boosted to around 32% with LotP and Mongoose, 1256 AP unbuffed).

Bad luck, small sample size, and using a level 60 instead of level 73 mob.
Your biggest shortcoming in your armory profile is your AP and to a lesser extent your crit; you have plenty of hit for your level of gear. I'm not necessarily saying you should replace all those expensive +8 hit Dawnstones, but based on all the theorycrafting going on here, you need to shift your focus. Apart from that, doing good damage as enhancement is relatively straight forward, ensuring we have plenty of free GCD time to drop our totems. Get WFWatcher if you don't have it, to help you decide when to Stormstrike; get a mod with totem timers if you don't have one (I use Yata). You note that you had Kings; the most important buffs you can get from others are Battle Shout, LotP, Salv, and Might, so hopefully you had all those too. One final thing, my damage on this week's Gruul sucked because I had just gotten a new arena Cleaver and my Axes skill wasn't maxed yet -- was your Maces skill maxed? I saw a short-term decrease, but once I got out to the Blasted Lands and maxed my weapon skill, I can now see (and it's reflected in SWStats/WWS) that with the arena weapon OH I am doing significantly better DPS than I was using a 71.7 DPS blue; I'm excited to get one for my MH in ~5 weeks.

Slightly OT..

Last night the raid leader "chopped [my] group's balls off" by swapping out a feral druid and a rogue for an OT war+warlock/imp so the OT could get my resist totems (cycling WF/NR while learning Hydross.) We had just come from Gruul where my group was me/fury war/feral/rogue/rogue and we wrecked house.

I'm kind of torn.. it seems silly to make those changes to add a measly 70 FR/NR to an OT. Splitting up the melee cost us a lot of DPS. I understand this class has a lot of "utility" but is this sort of group switching optimal, or at least to be expected, or was this just a bad move?

We're a casual raiding guild, (twice a week,) and I just switched to enh after gathering some gear. The raid leader is a smart guy, but everyone is still trying to adjust to whats best in terms of shamans and grouping as they're relatively new to us (alliance.)

(We currently have two shamans, one resto and myself [enh.])


On a tangent, I'd appreciate some critique on my gear choice/gemming if anyone is bored. Here is a link to my armory. I raid with Hourglass instead of the PvP trinket. I know I need a pants upgrade badly. I'm just torn on if gemming for +hit is really worth it when I'm sacrificing so much AP.

Thanks.

edit: Also considering dropping LW (was leveling it for the ele gear before the CC nerf) for Eng to get the helm. It ranked very highly on lootzor using the values that have been posted here. Worth it? (For being in a casual guild that probably won't see BT for months.)

The helm is once piece of gear, and T4 is pretty nice.

LW gives you 6 pieces of armor that are either better than T4, or not-covered by T4 (belt/bracers) and are very nice with sockets.

Personally, I am sticking with LW. Working on my Netherstrike BP now (5 more airs!).

Don't you guys have a hunter who can put up AotW for NR phase? Bad move, IMO!
Everything in this thread thus far indicates +8 STR gems to be superior to +8 hit gems, by a significant amount. If you want socket bonuses, the math indicates you should be dropping +4 STR/+4 Crit into yellow, +8 STR into red, and +4 STR/+6 STA into blue; but do note that a lot of the socket bonuses aren't worth giving up just +8 STR for. As far as pants, go run some Heroic SP, it's an easy heroic and the last boss is great for a shaman (heroism + poison cleansing totem); get the Midnight Legguards and socket with +8 STR, +8 STR, +4 STR/+6 STA. Your upgrades from there are the Skulker's Greaves from Kara, the Cataclysm Legplates from Karathress, or the Void Reaver Greaves from... Void Reaver.
Well, the engineering goggles take longer to go obsolete than the Ebon Netherscale gear, but Cataclysm Helm (token drops from Lady Vashj) is absolutely comparable, and neither the Cyclone Helm nor a Malefic Mask of the Shadows from Gruul are that far off from Surestrike Goggles 2.0. So, I wouldn't do it -- if I did anything, it would be to pick up Herbalism so I could farm flasks more easily. My comment may be biased by the fact I'm still saving for my epic flying mount, though.

I believe the values were the average from a sim over a range of ~1200-2400 AP, ~18-32 crit, and ~18-32 hit (including talents). If you go back about 15 pages you can check for yourself. As you stack more of one stat, the others become more valuable, so that is something to keep in mind.

Oh, one thing I forgot to mention, when you get your new set of legs, be sure to put Cobrahide on instead of Clefthoof like you have now.

This might have been answered in the last 50 pages but I feel the need to ask it anyway.

Is Relentless Assault always the best raid buff to have or are there better elixir combos?

The flask is always the best. Also, the +8 hit gem is never the correct choice. If you made a list of choices and ranked them in order of correctness, it wouldn't even be in the top 3. +8 strength, +4 strength/+4 crit and +4 strength/+6 sta are the correct choices.

Yes, i assume an even or better a normal distribution, in fact i assume that every windfury testchain will converge against the expected value of the probability, when the test chain is long enough(law of great numbers). And in fact i use the LaPlace Probaility, as seen here:

Pwf=1/(X+N). LaPlace says that every probability p is equal to k/n where n is the total amount of tests, and k are the tests with positive result. The only thing what i do is saying that k is 1 and n=X+N.

And as i said in the european forums you can transform every wf testchain into a normal destribution with p=1/X, by substracting N*k from n, assuming that the appearance of N strikes without windfury chance are part of the event, when windfuror procs.


In fact, I used floor before i made the simulations. And i don't see the point, why this end function should be continuous, when it simply is not. LaPlace Probability doesn't prohibit that, nor does any of my other assumptions do. What i say: the wf chance pWf is a laplace propability, and a function pWf(Wm/Wo) is not continuous. Well, you might say now, that we need a continuous model when using flurry, because it is not only up, so here you get it:

pWfFlurry=pFlurry*pWf(Wm/1.3,Wo/1.3)+(1-pFlurry)*pWf(Wm,Wo)
where pFlurry is the chance that a single attack is flurried. And you will see, that this function is continuous for the pFlurry parameter.

Well, my newest date says, that slow weapons are a bit better, but between 2.2/2.7 weapons and 2.7/2.7 is a less than 1% difference. This difference is gone in the moment, where you only got a 2.6 offhand, thant both combinations are equal. But in fact this whole thing is extremly depending on equipment. slight changes in hitchance or ap bonus will change the whole thing.

(Btw: i didnt write the thing with the damage bonus for fast/slow weapon combinations, because i forgot it , for the ones who are interested in what we are talking about: i said, that a fast mainhand and a slow offhand are better for wf, since an offhand proc does less damage, so getting more wf procs in the mainhand is preferable, in fact the factor looks like thatamMod=(PM+(1-PM)/2)/0.75. the factor 2 isn't correct, since the wf bonus doesn't get the offhand malus, but the correct value has to be calculated for each character separate since it is equipment dependent)

And to the 56.7%:
When we wield a fast and a slow weapon, the fast weapon swing more often than the slow weapon. So, when a weapons swings more often, it also gets more chances for a wf proc. Since the chance for a wf proc without the 3second timer is p=1/3 and has to obay to the law of big numbers, the fast weapon has to get more wf procs.

Or a bittle less abstract: we wield a 1.5 and a 3.0 weapon. so, when the slow weapons attacks , it gets 1 wf chance, but in the same time, the fast weapon gets 2 chances. Though we are not in the 3second wf cooldown, each attack has the same chance to proc, and thus the slow weapon has a 66% chance to proc, while the 3.0 weapon only has a 33% chance.

A hunter/pally works as well for NR/FrR. Do you not have many hunters/pallies, or are they in stacked groups?

If we actually assume normal distribution for WF (which I still think is invalid atleast in parts, the avg proctime between let's say 3.1 and 3.2s will be certainly normally distributed). Anyway I have some certain problems to fit a gauss curve to a 0% for <=3s to something unknown for >3s function, if we take the time difference for the swings into account.

That's why I said markov process and not bernoulli process, in either way, your flurry formula is correct nevertheless.


As for now and dual wield, it is probably the best way to assume that WF procs every (normal distributed 3.1-3.2s) and calculate the appropriate WF proc chance for each weapon using flurry values and use some weighting to distribute the wf procs properly for the MH and OH.

The problem I just see with your approach is, that your base assumptions like even distribution are not really true, that's why I suggested the use of markov-chains

And yes, we already know the enh shaman isn't easy to predict for equipment changes, that's why I wrote my tool for it.