Nisall

Don't take the percentages from wws to literally since they are whole intergers.

calculating the percentages by hand;
wws part 1
Hits = 11275 (53.36%)
Crits = 1738 ( 8.23%)
Glancing = 5163 (24.43%)
Miss = 1632 (7.72%)
Dodge = 1322 ( 6.26%)

wws part 2
Hits = 4627 (54.08%)
crits = 677 (7.91%)
glancing = 2017 (23.57%)
miss = 690 (8.06%)
dodge = 545 (6.37%)

As you can see they are a lot closers together than the site would imply.

But it was all done in one test so the numbers should be
hits = 15902 (53.57%)
crits = 2415 (8.14%)
glancing = 7180 (24.19%)
miss = 2322 (7.82%)
dodge = 1867 (6.29%)

Paperdoll crit 13.03%-8.14%=4.89% crit reduction

So looking at the 30k white hits all the values are what we would've expected them to be including the crit reduction

-------------------------------------------------------------------------------------------------------
Crusader Strike
overall data
hits = 6077 (78.48%)
crit = 522 (6.74%)
miss = 638 (8.24%)
dodge = 506 (6.53%)

paperdoll crit 13.03%-6.74%= 6.29% crit reduction vs bosses

 

Nisall

Yes, I'm getting 8% from talents.

You can see all the number on wws. Looking the columns add Landed (this already includes glancing) + crit + All misses (includes misses and dodges) to get the total number of swings taken.

Isn't this the 2-roll system? so this would be correct for the CS portion, but white hits are based on the 1 roll system.

Mmm... that reminds me I didn't calculate the CS part correctly

522 crit/ 6599 (hits+crit) = 7.91% crit

Paperdoll 13.03% - 7.91% = 5.12%% crit reduction

 

Valca

Are we still stuck to the "prioritizing system+FCFS" or is there an fixed rotation possible?
I dont know how you play but i dont like this prioritizing system.
1. It's totally depending on ping/fps in my opinion
2. Watching your bars makes "me" immobile or at least its harder to do top dps with a prioritizing system.
regards Valca

 

Amphi

According to info from the class forums, the Seal of the Martyr is now working as intended on the PTR.
This would indeed be good news. Anyone able to confirm?

Edit - Source: https://forums.wow-europe.com/thread...01790309&sid=1

Doh! Stupid proxy here at work caching sites forever. There was 2 new pages for me to read that didnt show up

 

Nisall

Amphi, that has been confirmed here on the previous page.

 

Maylander

With the buff to Seals, you should notice a significant dps increase unless your current prio button mashing is ideal, since so much more of the damage is guaranteed (as long as your hit/expertise is high). Beyond that, it's still a prio system I'm afraid, but CS/DS/Exco are better now (doesn't really change the prio on Exco, since it doesn't pop RV on targets). As far as I know, the new system is CS/DS/Judgement.

@Nisall
Hitting works like this:
1 Miss
2 Dodge
3 Parry
4 Block
5 Critical strike
7 Hit

So, basically, you get a test in this order:
1 Chance to hit
2 & 3 Expertise (back/front)
3 Bosses don't block
4 Critical %
5 Once you get past all that, you actually hit the boss

Edit: I should probably clarify my point. The misses in the WWS parse have to be deducted before calculating actual critical strike chance, since a "13% critical strike chance" in your melee tab only affects step 5 and 6 - it doesn't actually affect the 4 steps before it. 13% critical strike chance does not actually mean you have a 13% chance to get a critical strike, it means that IF you pass the other tests, THEN you have a 13% chance to get a critical strike. You also have to remove 3%, since this is a boss mob, landing you at 10% critical strike chance (if the hit passes the other tests, that is).

PS. I'm sure most people know all this already, but I think I failed to explain what I meant in my previous post, so I took the time to elaborate now.

Another edit:
The most accurate way of measuring would be:
- Get a lot of tanking gear with expertise/hit, but without agi/crit.
- Spec up to Crusader Strike without getting the crit talents, so it's only base crit chance.
- Perform the exact same test, this time with 0% misses of any kind.

This way, the output is so straightforward it's impossible to misunderstand. However, I think it's a lot to ask for someone to go to all that trouble just because we're curious about a mechanic.

 

Nisall

Unless I'm completely lost how the hits tables work I believe what you just explained is a 2 roll system. This is how yellow attacks work.

Auto attacks just use a one roll system. What you seen on your paperdoll is what you get, assuming equal level mobs. Here are 2 links which explain it in depth Wowwiki Attack Table and this thread started by Vulajin

 

Maylander

No, that's all in one roll, it simply goes through the steps one by one.

The two-roll system is something of a myth, trying to explain how certain attacks can get two effects at the same time (i.e block and crit, which is impossible in the one roll system). We don't actually know yet if the two-roll system exists, but we do know it's a bit different from regular hits.

Basically, the two-roll system is explained as the double of what I just explained. It means it goes through the stack twice (or perhaps just parts of it).

Like I said though, it might just be a myth. All we know is that there is a difference.

 

Exemplar

Someone correct me if I'm wrong, but we seem to be close on both white and specials to what has been calculated for Rogues in their thread.

Nisall's tests from Recount
Autoattack (one roll)
2475 crits / 30452 effective blows = 0.081 or 8.1% crit rate
Paperdoll 13.03 - 8.1 = 4.9%

Crusader Strike (two roll)
Rule out dodges/parries, only hits and crits are valid data.
6074 hits + 521 crits = 6595 effective blows.
521 crits / 6595 effective blows = .079 or 7.9% crit rate
Paperdoll 13.03 - 7.9 = 5.13% crit reduction

Glancing at the WWS there was ~1 min where Heart of the Crusader was up on the dummy (thereby increasing crit by 3% to all hitting the dummy, even if not in group) - this would affect the overall data in a minor fashion. 1 minute out of more than 12 hrs isn't much, but we're also testing fractions of a %.

These are, as I said, vaguely close to the 4.8% reduction derived from Rogues. The cleanest test at this point would be to engage at just below 4.8% crit and verify no crits occur, then retest just above 4.8% and prove a crit occurs (as Rogues did).

The problem is that a naked paladin has 3.27% base crit chance + bonus crit from agi. The race with lowest agi is a dwarf at 86 agi. 3.27% crit + 86 agi worth of crit comes to ~4.91% crit. It is therefore impossible to cleanly and definitively prove that you cannot crit below 4.8% crit rating.

I suggest some naked testing of white and CS. If there are crits we know the reduction must be lower than ~4.91% - we should be able to get some estimate on whether it's .1% or greater. This also would allow easy testing of Heart of the Crusader. Going from .1% or .2% crit (after reduction) to 3.1% or 3.2% is a huge swing.

Recommendation:
1) Naked test with only 10 Talent Points (Benediction, Imp Judgement, Heart of Crusader, no deeper talents which affect crit or Judgement crit) to test whether/how much Heart increases white and Judgement. Just autoattack and Judge. We should easily be able to determine if it is adding 0% or 3% to one or both.
2) Naked test with CS but no +crit talents (no Sacred Seals/Sanctity of Battle, Conviction, etc). This will get very low crit rate white and CS data. Here's a possible build - avoid Vindication (just in case) and all talents which would affect CS or white crits.

Rock: "We're sub-standard DPS. Nerf Paper, Scissors are fine."
Paper: "OMG, WTF, Scissors!"
Scissors: "Rock is OP and Paper are QQers. We need PvP buffs."

 

whistler-z

A 1-roll system has no steps but 1. In a 1-roll system, your chance to crit is not reduced by your chance to miss/dodge/parry/etc. (unless your crit is higher than your chance to hit (100-combined miss), as the crit fills in the hit area of the chart).

In a 2-roll system, you check to see if you hit before checking to see if you crit (simplified, but you get the drift). For instance, with a 10% combined miss and 30% crit rate, on average a 2-roll attack will only crit 30% of (100-10%), or 27% of the time. This is why DPS casters typically have such a huge emphasis on hit, as they start off with a base 17% chance to miss against bosses, and therefore also lose 17% of their crit.

EDIT to try to add something of value to the actual thread:

We don't have any reason to suspect a difference in our crit reductions between Live and PTR do we? I ask because running both clients at the same time allows more data to be collected at once. Don't know that both clients can take advantage of the macro-enabled keyboards, but at the very least additional data about white hits can be gathered simultaneously. I can collect some additional white hit data myself this weekend, but I can't significantly contribute to the yellow hit data until next week (ordered a G13 to play around with).

 

Exemplar

Anyone who has the time and connection could test data. I haven't only because I keep forgetting and I'm paranoid about unattended playing, even vs target dummy. If you do not have a keyboard which supports programmable macros, you simply use a program like Autohotkey (which last I checked has not been made illegal by Blizzard and is in heavy use by the multi-box community).

Setting a macro to perform multiple commands, time-delayed, at the press of a single button has always technically been illegal according to Blizzard. However, they have apparently been lenient in edge cases - tests on PTR, or tests on target dummies. I have yet to hear someone temporarily or permanently banned for using such, even back when people tested on Blasted Lands invulnerable mobs, but it's possible that it has occurred. They appear to passively allow such usage for testing purposes, whereas if it were used for exploitation (farming or similar) then it would incur negative repercussions.

Caveat Emptor, however. Just because I haven't heard of it or they haven't really done it in the past doesn't mean they cannot start applying penalties to those who use a macro keyboard or macro program, even for testing.

As for those "without WWS" - it's not a mod. Just turn on the combat log (/combatlog). When you're done testing, log out. Go to the WWS website, sign up for a free account (yes, free) and follow instructions. Recount is a mod, but won't show external info like other people debuffing your target.

Rock: "We're sub-standard DPS. Nerf Paper, Scissors are fine."
Paper: "OMG, WTF, Scissors!"
Scissors: "Rock is OP and Paper are QQers. We need PvP buffs."

 

Durinix

Ok, this is how the statistics work.

There is a probability p that a crit event occurs from n attacks which is measured by someone smacking a target dummy for a long time.

For a melee (white) attack, the number of attacks is given by
WWS lists the sum of hits and glances in the first column.

For crusader strike and other special attacks (as they are a two roll system), the number of attacks is
The observed crit rate is then given by:
There is a statistical error in the observed crit rating. If there is enough data, then you can approximate the data to a normal distribution. The required condition for this approximation to be good is that
By approximating the data to a normal distribution, you can calculate the range of values for which are within a margin of error.

Where z is a really complicated number based on the percentage of possible results which you want to have lie in the bounds. I've been stating a standard error (1 standard deviation from the mean) which includes ~68% of the possible population.

When expressing a value, the error is represented like so:
All of this means that the ideal error is when you have a 50% chance to crit.

The reason why we don't run in parallel between live and PTR is that you're assuming that there isn't a difference. There's no reason to add that assumption. It just complicates the matter more.

Even 1 min of Heart of the Crusader or any crit increasing buff is enough to invalidate your data. One of the assumptions you're making is that your crit rate is the same over the entire simulation. We can't make any sort of error calculation that can take that 1 min into account. If you want delete that minute, and we can use the data but otherwise, it's useless.

If we could get a 4.8% crit chance while naked, then making 10000 attacks without a crit actually proves nothing. You can not pin down the exact crit rate in such a fashion. Once you get to that kind of chance (nearly zero to zero) almost all statistical models break down (and the ones that don't break down can't deal with that kind of data because they make calculations like the factorial of 10000).

Talented crit which does not apply to a particular ability is listed in the paper doll crit. It doesn't matter if you take the talents or not, just as long as you SS the paper doll.

Anyway on to data.

Nisall's data (melee)

Crit -> 8.14 ± 0.16%
Glance -> 24.19 ± 0.24%
Miss -> 7.82 ± 0.15%
Dodge -> 6.29 ± 0.14%
Crit Reduction -> 4.89 ± 0.16 %

Crusader Strike data
Crit -> 7.91 ± 0.32%
Crit reduction -> 5.12 ± 0.32%

Greatrichie's data (melee) (PM data WWS here, same gear as before)
Crit -> 29.36 ± 0.13%
Crit reduction -> 5.19 ± 0.13%

CS
crit -> 29.76 ± 0.2 %
crit reduction -> 4.79 ± 0.2%

Chi squared analysis with a hypothesis of a 4.8% crit reduction on all WWS data for melee attacks does not reject the data (for tampering) and does not reject the hypothesis (for statistical significance).

Chi squared analysis with a hypothesis of a 4.8% crit reduction on all WWS data for Crusader strike fails 10% significance (ie, there is only a 10% chance of getting data that is worse that what we observed if the hypothesis is true). This isn't enough to reject 4.8% crit reduction on CS.

Looking at some of the runs, one of the runs included HotC. This could be what is causing this issue. Removing that dataset (for the possibility of there being an issue with HotC), the chi squared analysis on Crusader strike does not reject the data (for tampering) and does not reject the hypothesis (for statistical significance).

I'd say that we can be fairly certain that we have a crit reduction of 4.8% against boss mobs for both white damage and special attacks.

 

Endoscient

If all specials use a 2 roll system, it would dramatically increase the value of hit and expertise. I just implemented specials using a 2 roll system in Rawr to get an idea of the exact numbers. Here are some comparisons from a character with BIS T7 level gear, and full raid buffs.

Stat	2 Roll	1 Roll
Hit	2.23	1.72
Strength	1.64	1.62
Expertise	1.62	1.29
Critical Strike	0.90	0.91
Agility	0.85	0.85
Haste	0.77	0.74
Armor Penetration	0.71	0.68
Attack Power	0.65	0.64
Spell Power	0.19	0.19
Intellect	0.01	0.01

Note: These numbers are just used to compare the value of hit in one example gear set, and there value change based on your gear. Please use a theorycrafting tool for numbers accurate to your character.

 

• frmorrison

SS doesn't proc on Blood hits, but rest assured that every boss encounter in Ulduar involves AoE (even the vehicle boss), so you will get lots of shield procs.

Is Rawr taking into the 2 rolls of our two spell attacks (every tic of Cons and Exo) with these numbers?

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Endoscient

Yes I did for Exorcism, but Consecrate can't crit so it doesn't change it.

 

Exemplar

I implemented the same in Bellator's before lunch (not yet available for download - maybe early next week), but it took discussions with someone else to point out what Ermad saw immediately. Hit and Expertise scale with gear. The better your gear (i.e. the harder every attack hits and the more you should crit), the more important Hit and WE become. Not hitting means you cannot crit, so Hit and WE "steal" some of Crit/Agi's effective value. Both Hit and WE eventually surpass Str as most important stat. Ermad, I assume you're seeing some gear levels where this occurs as well.

Examining the Crusader Strike data so far, I'm forced to think that our specials are indeed two roll. In a one roll model crit is almost a full 1% higher than expected, while two roll it matches what we'd expect (see Durinix's post above).

This means we may no longer be saying Str is always best. Indeed, in high levels of gear it may be that we will socket Hit/WE if below cap. It's also probably advantageous for Alliance to have two gear sets (like casters in case no Spriest/Boomkin) - one for when you have your Draenei aura, one without.

Ditto on the not for Consecrate thing, same reasoning.

I think Bellator's, Rawr, and Redcape are all approaching convergence in regards to output.

P.S. If we actually want some Hit/WE, that means more socket bonuses for the non-JCers!

Rock: "We're sub-standard DPS. Nerf Paper, Scissors are fine."
Paper: "OMG, WTF, Scissors!"
Scissors: "Rock is OP and Paper are QQers. We need PvP buffs."

 

Marxalian

I have been messing around with the beta version this morning and I must say, bravo on making it a bit more solid and more user friendly.

I do have a question regarding the relative stat weight function. I understand that hit/exp will be inflated when under cap, but are the changes to haste really enough to have it outweigh strength? I am currently getting a value of 2.17 for haste and 1.66 for strength. I have a fairly high level gear set so my first impression is that maybe i have reached an ap cap where other stats surpass its value.

I know that there were disputes with stat weight between rawr and redcape and to my understanding that had been corrected. So, is this to be taken as correct and that haste, in my case, is a more efficient way to increase my output?

EDIT: appologies, meant to post this in the rawr thread

 

Exemplar

Think I forgot to mention (couldn't find it on a search), but with JoB damage reduced (SoB increased to balance) when under 20% Hammer of Wrath is downright superior. So when everything is off cooldown under 20%, HoW is your first priority.

Now it's HoW > everything else.
Previously it was Judge > HoW > everything else.

Rock: "We're sub-standard DPS. Nerf Paper, Scissors are fine."
Paper: "OMG, WTF, Scissors!"
Scissors: "Rock is OP and Paper are QQers. We need PvP buffs."

 

Dahkness

Long time reader, first time poster. Not to detract from the current topic, but I was curious as to whether or not anyone on the PTR has tried this spec:

http://talent.mmo-champion.com/?pala...0&version=9742

The way I see it, most pally's look to be going down the Prot tree to get DG. Sacred Shield is nice and all, however, it's 12% base mana and negates approximately 1500 damage with most end game ret gear?

I was looking at 14 in holy. First, if what I have seen is correct, 5/5 SotP is >> SoC (When SotM isn't viable...don't know the Ulduar fights as of yet). Secondly, 12% more healing from FoL for only 7% base mana coupled with 4/5 Illumination giving you 80% chance to give you back 60% mana on crits.

If I am over looking anything please let me know as it looks as if I may certainly go 14/5/52 come 3.10.

 

Telumehtar

I've been reading and re-reading some of the earlier arguments for glyph of blood, vs glyph of consecration and I think I've finally been able to formulate my hesitation with considering Glyph of Consecration as being the better glyph.

Under theoretically perfect conditions, I can see where the mana savings in conjunction with the reduction in conflicts would pull Consecration ahead in optimal "stand and deliver" style fights. But on fights that require more mobility, and in which consecration is going to be a devil to maintain on a target (Heigan would be a prime example), I'm not so sure that consecrate keep's it's place.

If I understand the arguments properly, Glyph of Consecration's mana savings are definitely behind the mana regen produced from Glyph of Blood. But it's the GCD conflict avoidance which gains it a favorable place over Blood. However if my thinking is proper, the moment you lose 2 ticks on a Glyphed Consecrate you've essentially squandered the mana savings and DPS gain that the glyph provides. Given that Glyph of Blood has fewer issues with mobility and if some of the Ulduar fights are shaping up to be a little more mobile than your classic Patchwerk, does Glyph of Consecration still hold it's place in our three glyph choice?

 

• frmorrison

You are missing a bit. DG is a powerful "healing" tool, better than FoL by a long shot. You also can buff it with 17 points in Prot.
Remember SS may only absorb 1200 damage every 6 seconds however it lasts 30 seconds (60s with 17 in Prot).

Also, getting Illumination 60% mana back from the few Flashes you will use isn't much mana, 4.2% only 80% of the time, so basically halves the low cost of Flash.

However, that may be a good single spec to heal heroics and do Ret dps in raids, but that point is moot with spec swapping (assuming Holy was your second spec).

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Dahkness

@frmorrison

Ah thanks, my brain had a slight misfire for a second thinking that SS had to be reapplied every time it absorbed the damage. Thanks for correcting me.

 

J1M

The real DG bonus to me seems to be a self-shield that will last for 1 min. Certainly help solve the SoB problem. However, 12 or 13 points in holy could be interesting, but I would take improved LoH over illumination.

 

beta4Life

Remember it is 1500 damage every 6 seconds for a minute, if used to its full extent (probably won't happen much) you are essentially getting 15k healing for your 450ish mana, plus the added bonus of essentially guaranteed FoL crits on yourself at all times.

In my experience in Ulduar with the new DG (Hodir hard and Iron Council Hard), the shield is huge, it was absorbing essentially all of my SoB recoil and then a little extra (bugged SoB, so might just cover recoil now). In most fights I was actually dead last in damage taken, even though I damage myself constantly to DPS.

After my experiences and assuming nothing changes between now and 3.1 going live, 0/17/54 will be my spec.

 

flexbutt

The other problem with that spec is you didn't take PoJ. If you plan on progression raiding, then 15% move speed should be a given. Death knights no longer have Unholy Aura, so getting the talent is 100% justified.

The overall objective is to be able to improve survivability without lowering dps. A potent Sacred Shield that lasts a minute, in addition to an extra powerful JoL from Divinity will end up costing much less mana and GCDs than flashing yourself and hoping it will crit for an 80% chance of an Illumination proc. As such, http://talent.mmo-champion.com/?pala...5&version=9757 will be the most viable, give or take some filler points in prot.