Ok finally got around to editing this post. It started when I noticed a consistent lack of (particularly) shred crits over an extended period, at first I just put it down to luck but after a while it got to the point where I started to think something must be wrong. I started noticing this back last June in SSC, but only thought to investigate it further at the start of December, all data collected thus far shows a significant lack in crit%.

First things first, the mechanics of critting:

• Crits are based on a random number generator, and are inherently luck based. This means that even if you crit seventeen times in a row, the next attack still has the same likelihood to crit, and vice versa. However over a large enough sample, the values -should- even out to the pre-determined value
• Only attacks that actually land have the potential to crit. This means that if you miss 6% and a further 4% are dodged, only 90% of your total attacks have the potential to crit. (Two-roll theory)
• A boss is 3 levels above you, this gives -0.6% crit chance (-0.2% per level), assuming the npc defence ratings work the same way as player ratings.

The “correct” formula (the one you get from theory crafting) for determining whether any given attack will crit on a boss is:

Crit chance = ((1 - A – B – C – D) * X)) – 0.006

Where:
A = Boss parry chance
B = Glancing blow chance
C = Boss dodge chance
D = Player miss chance
X = crit chance on character sheet
(1 = 100%, 0 = 0%)

Due to this, there has been a large amount of confusion and needless discussion thus far. In order to calculate the actual crit% of shreds that landed, you can use this simple formula:

X / Y = Z

Where:
X = Total number of critical shreds
Y = Total number of shreds that landed
Z = Shred crit%

This immediately removes any discussion of misses/dodges/etc, and any numbers you see quoted by Moanique or me use this formula.

Crit% soft cap
It has been suggested this is due to the soft cap in crits. This theory says that you cannot crit more often than you hit (simple stuff). However it really only applies to melee attacks from the front, which can be parried, dodged, miss and glance, adding up to a very large % of attacks that cannot crit – this is not applicable to shred since it cannot be parried or dodged, until you reach around 85% crit chance

Ok, now that the mechanics has been covered, on to some actual data:

NB: Data over the thread is a bit spread out, with people referring to data from page 2 on page 7, and so on. I’ll try to get all the data summated in discrete blocks. Also, many of the WWS reports have expired, so if you didn’t check them before they wont be viewable unfortunately.

ALSO!! It is important to note that bosses defense level means they take 0.6% less crits, I have not included this when quoting crit% however I have used it while performing statistical tests and the results remain unchanged.

Toskk, independently collected data linked here:
http://druid.wikispaces.com/ToskksDPSData

Serpenstar: 7th January data
(see post #10, page 1),
Data collected over a 4 hour raid, unfortunately the WWS reports have expired and I didn’t think to save the data.

Post 74, page 4:
From a sample of 222 shreds, 85 crit. This gives a 38.28% crit rate.
45.6 - 38.38 = 7.32% discrepancy.


Moanique: 8th January data
Post #47, page 3:
Shred:
hit: 601 (59%) --> 32/(59+32)=36%
crit: 338 (32%)
miss: 78 (8%)

Mangle
hit: 278 (58%) --> 34/(58+34)=37%
crit: 165 (34%)
miss: 37 (8%)

post #72, page 4: I seem to be getting a pretty much precise 5% lower crit%, when I had 41% crit on my sheet I'd end up at 36%.

Both of these data samples were put through statistical tests and it was found that the crit was significantly lower than it should be with an accuracy of over 99%. This means that there is less than 1% chance that the lower crit% observed is due to luck!!

Serpenstar, 22nd January
See post 109, page 6
Again, unfortunately, the WWS reports have expired. I confirmed this data by keeping Recount running for the duration of the raid, which gave identical results.

Total #Specials that landed - 789
Total #Special crits - 299
Total #Special hits - 490
*Total specials crit% - 37.90

Predicted/Expected crit% (A) = 44.07
Crit% from data (B) = 37.90%
A - B = 6.17%
Sample Size = 789 hits

This sample also falls over 99% confidence, there is less than 1% chance that a crit% deficiency of 6.17% is due to bad luck on a sample of this size.

Moanique, 1st February:
See post 153, page 8
http://rapidshare.de/files/38460926/yellow.xls.html

Updated spreadsheet, again showing a deficiency in shred crit%.

Serpenstar, 2nd February:
See post 158, page 8
I used /combatlog then filtered it and went through using “pen and paper” to count the number of shreds that hit and shreds that crit, these were the results:
Total shreds: 948
Hits: 534
Crits: 356
Dodges: 44
Misses: 13
Parries: 1 (lag?)

This gives a discrepancy of 6.03%, of shreds that landed. This, again, is outside the suggested confidence interval for 99% accuracy.

Serpenstar, 14th February:
See post 200,page 11

WWS links:
http://wowwebstats.com/fff2315v5lmac
http://wowwebstats.com/4gzynbbyohfku
http://wowwebstats.com/zpz325brhone3
This data was also confirmed by Recount, and /combatlog data. Unfortunately I do not have links for these.

Total Shred hits: 453
Total Shred non-crit hits: 273
Total Shred crits: 180
Shred crit%: 39.74
Deviation from norm*: 46.03 - 39.74 = 6.29

AGAIN this data is 99% accurate, there is less than 1% chance that the deficiency in crit% is due to chance.

NEW DATA:
1st May 20o8, post 345, page 18.

Serp vs Brutallus...

http://wowwebstats.com/lqribrm5rp1o5?s=6735-6821

Char sheet crit chance: 51%

Total shreds: 674
normal hits: 377
crits: 297
crit chance: 44.06%

~7% discrepancy again.


Customer support forums thread:
http://forums.wow-europe.com/thread.html?topicId=2405193781&postId=24049590817&sid=1#5
locked with no suitable answer or explanation.
http://img205.imageshack.us/img205/4537/20080128io7.gif

Apologies for the wall of text, tried to format and edit it as much as possible to make it easier on the eyes.

Total number of samples = 5607

Every single sample of data collected showed a large and significant decrease in crit% from what it should be with an accuracy of over 99%. These statistics are far in excess of the proof required for any scientifically measured data.

Thus it is proven that the shred crit% consistently is around 5-6% lower than it should be, the chance that this is due to “bad luck” I would estimate around 0.01%, however statistical tools limit me to saying 1%.

It's been suggested that lotp might be bugged, as the 5% from this talent seems close to the numbers we've noticed.
I'm currently trying to collect some more data for this by looking at crit% of other classes on raid bosses with and without lotp - as you can imagine it's pretty hard getting enough samples as I need to know everyone exact crit%, they have mongoose procs, different buffs, and so forth..

NB. LoTP has been ruled out as affecting white crit% through testing. Sharpened claws or other game mechanics may be at fault.

ur crit is afected by ur adversarie defense, a boss has 3 levels higher than u, wich gives him some defense.

It is still not enough to justify those values u are giving us, but i guess it is one of the reasons, have no idea what other factors may be causing that.

PS: oh i just thought of 1 more possible factor, maybe the addon ur using to get ur % values counts misses as "hits", wich ends up giving u wrong values on ur actual crit %, however if u have good hit rating this is yet another very minor factor :)

Aye thats what i thought, but it shouldnt be enough to consistently lower crit% by at least 5%, and often by 10%.

It's extremely annoying as i've built/enchanted/gemmed all my gear towards building up as much crit as possible and its not doing what it should be

Maybe some kind of cap.. u should try reducing ur crit rate slightly and using the same addon, if the values remain the same then u just found a hidden cap.

Maybe the bosses are wearing resilience gear these days?

But no that wouldn't make sense. Since you said other classes don't seem to suffer from it?
Probably just your luck then, would like some more info on how other classes are affected (without mongoose).

I went looking for some theary craft pages on wow wikki when i noticed my crit chance on special abilitys seems pritty low. I only have about 30% crit anyway (which isn't ammazying) but i seem to almost NEVER crit with FB which makes me stop using it, and shred is about 2/12, usually incrased if the player is a cloth wearer. Most of the time i crit when i really dont need too.

Most annouying is that FB with a full energy bar can do 3500 damage for me on cloth.. and maybe 2.5-3k on plate/heavy armour classes. But when i wait for 100 energy it crits 1/15.. which is really frustrating.

i think it all drops down to the lvl of the mob that you are fighting if your crit chance it 47% for example, the boss is 3 lvl's higher than you so its defence is incresed by about 15 points so thats 345 defence (basic) that will drop your crit chance by say 1% so thats 46% left. Crit chance is on All attacks not just yellow damage if your normal Hit was a crit b4 u used Shred then chances are your Shred wont crit.

Basicly stop thinking of Crit as Yellow damage only check what it is with your white crits aswell and then that will give you a better idea, ( the defence and armor and all that crap on the mob does effect it alot )

What is your miss rate? Yellow attacks calculate crit after it is determined wether it hits or not (two roll system), therefore if you miss 10% of your attacks, you get only your crit chance of the remaining 90%. Say, 40% of 90% is 36%. Include the boss' defense reduction of crit and your 40% is easily turned into 35% and even less.

White attacks use a single roll system, which results in the slightly higher crit chance you mentioned.

---

Q u o t e:
Crit chance is on All attacks not just yellow damage if your normal Hit was a crit b4 u used Shred then chances are your Shred wont crit.

---

The result of the previous attack doesn't affect the income of the next attack. The only thing which it matters for and what you can count is the meaningless cumulative chance to crit.

---

Q u o t e:

Basicly stop thinking of Crit as Yellow damage only check what it is with your white crits aswell and then that will give you a better idea, ( the defence and armor and all that crap on the mob does effect it alot )

---

That's probably the crux of it..

It depends on their mechanics, on whether the engine runs on some kind of true random mechanic or a compensation technique.

(disclaimer: I'm not a software engineer)

The way I imagine it, it has one of two:

A "true" random engine, that can lean towards more or less probable outcome, allowing really strange end results some times. Like killing 500 EPL larva and not getting a single larval acid, or killing 10 and getting one on each kill, the day after.
This could be true, that the random engine is based on a multitude of things, from how many people are online, what date it is, how many horde wins in WSG to the balance of item quality worn throughout the server, mixed up in a confusing algorithm..

the other is a compensation engine, that adjusts your actual crit-rate in comparison to the estimated kill time of the mob you're fighting. Or, DPS versus mob HP.
Meaning that if you've been unlucky on the first 200 hits, you crit almost every time on the last 100. (edit: provided your crit chance is 30%)


If it's based on the first, then there's the chance that you can be just plain unlucky.

If it's based on the second, then it's possible that it's the crit % of your total attacks, and not just each individual move.

In the latter case, I figure we're kind of pre-nerfed since white damage makes up most of our number of attacks. In that case, our yellow moves have inherent less chance of critting.

looked up some webstats reports for it, just from the last raid, just linking some of the fights when i was dpsing when i was dpsing - dont look at the whole raid because i healed some bosses and tanked at others (hybrid dream eh? :P)

(apologies for spamming links, put the crit% for shred next to links so dont click if you dont wish to)

RoS: I had agi totem in p1 (dodge for rogues and tanks), and wf for other phases. 46% crit without totem

http://wowwebstats.com/tte1nb1xo1vtk?s=15117-15218&a=8 35%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=6663-6827 22% (wtf?)

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=9253-9380 55% (p1 only, around 48-49% crit on character sheet - this was actually the closest % of all the tries to what it should be)

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=7344-7534 39%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=8044-8187 36%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=8314-8473 37%

Mother Shahraz: no agi totem, and dps flask used over elixirs, full shadow res equipped. 26% crit on character sheet

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=18485-18684 20%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=18920-19345 21%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=19656-19783 17%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=19957-20140 15%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=20334-20695 15%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=20965-21274 14%

http://wowwebstats.com/tte1nb1xo1vtk?a=8&s=21540-21936 25%

Should be fairly obvious to everyone that the crit% is 5-10% less for most tries, with only 1 try in the whole night over the crit% and only one other that is close to what it should be

And yes, i know its only one night but its not even close to what its supposed to be - feel free to look at other raids, i've replaced some gear over xmas and cant remember what buffs i would have had for them, and which fights i healed/tanked though. still should be fairly obvious im not critting what i should be

---

Q u o t e:
What is your miss rate? Yellow attacks calculate crit after it is determined wether it hits or not (two roll system), therefore if you miss 10% of your attacks, you get only your crit chance of the remaining 90%. Say, 40% of 90% is 36%. Include the boss' defense reduction of crit and your 40% is easily turned into 35% and even less.

White attacks use a single roll system, which results in the slightly higher crit chance you mentioned.

---

ah ok thanks, i wasnt sure if it was a two roll system or one roll - have read articls supporting both

linked some WWS in the post above so you can check there for all details - i have 8.6% +hit according to game UI, though some fights i missed alot/not at all as you might expect.

Next raid i'll look at hit:crit ratio only of attacks that landed, though i still think it wont be close. for example the RoS kill a total of 10% of all shreds missed, and i had a 35% critrate, if you include the two-roll system thats still only 38.5%

(i think WWS includes dodged attacks as misses thus causing elevated miss%, not sure though)

Svahva:
i believe that its based on a random number, and isnt "normalised" as you suggest - like you though im no software programmer so im not the expert. so how much white crit% there is should not affect yellows (and white crit% is still significantly lower than it should be)
and as i said, this isnt something i noticed one raid and thought "wtf!?!?!11", its something i've noticed consistently over months

If your crit chance is consistently low, night after night for weeks, then it's definately worth investigating.
But based off one, or even a few, it's nothing more than "random" =)

---

Q u o t e:
That's probably the crux of it..

It depends on their mechanics, on whether the engine runs on some kind of true random mechanic or a compensation technique.

(disclaimer: I'm not a software engineer)

The way I imagine it, it has one of two:

A "true" random engine, that can lean towards more or less probable outcome, allowing really strange end results some times. Like killing 500 EPL larva and not getting a single larval acid, or killing 10 and getting one on each kill, the day after.
This could be true, that the random engine is based on a multitude of things, from how many people are online, what date it is, how many horde wins in WSG to the balance of item quality worn throughout the server, mixed up in a confusing algorithm..

the other is a compensation engine, that adjusts your actual crit-rate in comparison to the estimated kill time of the mob you're fighting. Or, DPS versus mob HP.
Meaning that if you've been unlucky on the first 200 hits, you crit almost every time on the last 100. (edit: provided your crit chance is 30%)


If it's based on the first, then there's the chance that you can be just plain unlucky.

If it's based on the second, then it's possible that it's the crit % of your total attacks, and not just each individual move.

In the latter case, I figure we're kind of pre-nerfed since white damage makes up most of our number of attacks. In that case, our yellow moves have inherent less chance of critting.

---

No, no, no, no.

Which do you think is simpler for a programmer to code? A calculation of a stat from the items you wear or a robust, sci-fi bordering set of laws that dictate what happens when?

If you have a 40% chance to crit and you do a million hitting attacks out of which 20% are shreds/mangles, what percentage of those 200 000 yellows are crits? It's pretty close to that damn 40%.

---

Q u o t e:
No, no, no, no.

Which do you think is simpler for a programmer to code? A calculation of a stat from the items you wear or a robust, sci-fi bordering set of laws that dictate what happens when?

If you have a 40% chance to crit and you do a million hitting attacks out of which 20% are shreds/mangles, what percentage of those 200 000 yellows are crits? It's should be pretty close to that damn 40%.

---

fixed it for you :P

---

Q u o t e:
If your crit chance is consistently low, night after night for weeks, then it's definately worth investigating.
But based off one, or even a few, it's nothing more than "random" =)

---

yep, as im saying now for a third time, it IS consistently low night after night, for months

you can also look at the reports and look at Clarx's dps stats (he did 62% on the actual kill the lucky sod but the rest of tries were significantly lower, he had agi totem for the fight, so a higher crit% than mine)

---

Q u o t e:
(i think WWS includes dodged attacks as misses thus causing elevated miss%, not sure though)

---

Yes, it does. If you click on an ability like shred in WWS, you'll see the breakdown of your "misses", which can be dodges, parries or misses. On the couple WWS's I checked you didn't miss practically at all but there were some dodges and parries.

Also, on small sample sizes (like one pull) the weight of one crit or miss is too high.

---

Q u o t e:
No, no, no, no.

Which do you think is simpler for a programmer to code? A calculation of a stat from the items you wear or a robust, sci-fi bordering set of laws that dictate what happens when?

If you have a 40% chance to crit and you do a million hitting attacks out of which 20% are shreds/mangles, what percentage of those 200 000 yellows are crits? It's pretty close to that damn 40%.

---

Like I disclaimed, I'm not a software engineer, I don't know how the "If's" and "and's" work out in the end.
Who said it has to be simple?
Maybe they strived for an as realistic projection as possible, what do I know.

It's *plausible* that they just smacked something simple together.
I'm just giving theories as to how it might work out.

edit

it still doesn't clarify what's correct. Does the 40% crit apply to a total of your attacks versus the total of hitpoints? or is there a 40% chance of each of your attacks to crit?

There is a difference.

I've noticed the same in my logs but attributed it mainly to not being hit capped, attacking a higher level target and the way dodges and parries are counted. I never had the nerve to track it down. If were to I think I'd start by attacking same level mobs while being hit capped to factor out some of the uncertainties and to have more control over the experiment.

Moanique: Agree, hence me spamming every fight into forum :P, also the reason I didnt make this post months ago

Svahva: Each of your attacks has a 40% crit chance (assuming it lands). You can crit 1,00,000,000 times in a row and the next swing still has 40% to crit