Spirit and Crit Rates

[MongoJerry]

Background

According to Araxis, a rogue guildmate:

Supposedly, there's a theory going around in rogue forums that spirit reduces chance to crit. His claim:

200 swings on a priest friend fully geared showed a lot less crits than same priest friend who is naked. Being that the only thing that should affect crit is defense, it seems rather implausible. However, Caydiem has said on forums before "What if Spirit can reduce the chance to crit?"

200 swings seem awfully low though. So perhaps I will let it run for 1000 swings. Seems like an interesting change if not reported.

It seemed like this had a snowball's chance in hell of being true, but this is Blizzard we're talking about, so you never know. It would be a significant find if it were true in that raid groups would want to make sure to cast Divine Spirit warriors and rogues in addition to their mana-using members. It would also fit into the rogue mythos that Blizzard hates rogues, because you know, being able to ambush people for 2.5k non-crit just isn't enough. Araxis and I set out to test this theory thoroughly.



Methodology

We first tested Araxis's crit rate while I (as a priest) wore no spirit gear, so I had the base priest spirit of 130. In addition, I checked to make sure I didn't have any +dodge% or +agility items on. Araxis then attacked me from behind using only auto-attack. The tooltip indicated he should have a crit rate of 27.9%, and his weapon gives +3 dagger skill.

We then tested Araxis's crit rate after I put on spirit gear, which mostly consisted of Transcendence items. In addition, Jivox, a fellow priest, cast Divine Spirit on me and I used a Spirit of Zanza potion for an additional +50 spirit buff. (I also used a Gizzard Gum, but it turned out that it doesn't stack with the Spirit of Zanza). This gave me a total of 361 spirit. I again checked to make sure I didn't have any +dodge% or +agility items on, and I didn't wield a weapon just in case wielding one would create any wierdnesses with an unlisted parry or something.

Araxis used Damage Meter to tabulate the hits and crits during the tests, and we tested each case for 2,000 hits.

Data

When I had 130 spirit, Araxis reported crit rate of 32.5% after 2,000 hits. After only 1,000 hits, he had a crit rate of just over 32.0%, so the crit rate seemed to hover around this 32-33%ish range pretty consistently.

When I had 361 spirit, Araxis had a reported crit rate of 33.8% after 2,000 hits. At first, this may sound like a grand discovery that wearing spirit actually causes the crit rate against you to go up (Omg! Priests are screaming everywhere!), but this is likely due to some slightly skewed data. The test started off with an abnormally high crit rate. For the first 400-500 hits, Araxis's crit rate was on the order of 38%. For a while, we thought we might be on to some grand discovery. (Maybe spirit does affect crit rates but Blizzard got the sign wrong in the formula?). But as more data came in, the numbers began to normalize. By 1,098 hits, Araxis's reported crit rate was down to 34.5%, and it settled down to 33.8% by the end of the test. I suspect that if we continued for 10,000 hits, the crit rate probably would have settled in the 33%ish range.

Conclusions and Discussion

A target's spirit value does not affect one's crit rate to any visible degree. Too bad. No great discovery here.

The difference in the crit rate between that reported by Damage Meter (33%ish) and that reported by the tooltip (27.9%) can be explained by how the two perform their calculations. The tooltip crit rate is one's overall crit rate per swing (hits and misses) whereas Damage Meter reports crit rates only on hits. Araxis's miss rate was 13%, and dividing 27.9% by .87 gives a hit-only crit rate of 32%, which is approximately what we found. Thanks to Concillian at the Lurker Lounge along with information gleaned from the Elitist Jerks forum for this information.

[Concillian]

Now, the question I have for you all here is why is the crit rate reported by Damage Meter consistently 4-5% higher than that shown on the tooltip? Araxis has no talent spec that adds crit% to his white damage. It's possible that his +3 daggers skill is affecting things somewhat, but I don't think that would effect the crit% that much (if it did, watch for all rogues to suddenly start equiping +weapon skill items in PvP). I suspect that the difference is in how Blizzard calculates crit rates and how Damage Meter reports crit rates. Misses might be playing a role in one and not in the other, for example. Does anyone have a good explanation of the difference? I want to know in case we do some more tests and use Damage Meter to tabulate data again.

This could be the case. Supposedly the table lookup nature of WoW means that if you have a 30% crit rate you get 30% of all swings as crits regardless of if you miss.

Damagemeters could be recording it as a % of landed blows, not factoring misses into the crit rate. Assuming he has a miss rate of about 14%:
Then you might expect, on average 28 per 100 swings will be crits, 14 will be misses, and the remaining 58 normal hits.

WoW says 28% crit, which is 28 crits / 100 swings = 0.28 = 28%
Damagemeter could be calculating crit rate as 28 crits / 86 landed blows = 0.325 =32.5%

It's even within the realm of what you might expect for DW miss rates.

[MongoJerry]

We were discussing this kind of thing last night. That sounds right. Thanks for the confirmation. Araxis had a 13% miss rate, so that fits.

So does that mean that if you had a 100% crit rate, that you'd hit every time? Or does it check for misses first and then crits? Kind of wierd either way. It makes +crit% more valuable than it should be.

[Quark]

Way back when Tseric mentioned this was a possible change to spirit. Nothing came of it.

Edit: http://news.thottbot.com/index.php?authors...rch=spirit+crit

What if ...: ...spirit was given an attribute that worked as a counter to crit%? This would operate in a way that amassed spirit would serve to reduce the crit% of attacking foes, thereby reducing potential damage dealt and inversely affecting survivability. It might be something that takes effect solely in PvP.

Discuss.

Welcome to nerfville. Population: you and me if this gets implemented.

https://forums.worldofwarcraft.com/thread.a...ge&t=639665&p=1

You folks need to calm down about this.

Caster boosts for critical percentages are nowhere near the efficient formula melees have. It takes far more points for a 1% spell critical chance to be gained through Intellect than it does for Rogues through Agility.

This additional Spirit effect would most likely operate the same way: a lot of Spirit would be required to make any noticeable impact. As any caster can tell you, most items with a high amount of Spirit are significantly lacking in other qualities.

Again, it's not nearly as bad as you're making it out to be. Take a deep breath, sit back, and wait to try it out in testing before giving feedback on it. I really don't think it's going to affect your DPS as much as you're making it out to do.

[Concillian]

So does that mean that if you had a 100% crit rate, that you'd hit every time? Or does it check for misses first and then crits? Kind of wierd either way. It makes +crit% more valuable than it should be.

Uhh, I don't know. If you find a way to get 100% crit rate, let me know and I'll test it for you =).

If you want to verify this assumption, all you need to do is remove the off-hand weapon. The miss rate should drop to 0% without the DW penalty and the numbers between the character screen and damagemeters should get much closer.

Supposedly glancing blows can push hits off the table without enough +weapon skill as noted by the people doing that kind of research. So crit cannot push glancing blows off the table, the opposite happens... If you dual wield and have, say 30% crit and no +hit or +weapon skill, you will only get ~18% crit 0% hits and remainder glancing blows.

There's a reference thread on this in the Warrior forum, but the WoW forums are down (again?... still?), so I can't link to it.

[MongoJerry]

Uhh, I don't know. If you find a way to get 100% crit rate, let me know and I'll test it for you =).

If you want to verify this assumption, all you need to do is remove the off-hand weapon. The miss rate should drop to 0% without the DW penalty and the numbers between the character screen and damagemeters should get much closer.

Well, how about this question. One of the abilities of the Anubisath trash mobs between the Twin Emps and C'Thun is a to cast a dust cloud debuff that reduces your chance to hit by 85%. Under this theory, every hit should be a crit if you have more than a 15% crit rate. But rogues and warriors do get non-crits, so what's causing this?

This could be the case. *Very helpful info snipped*

I went ahead and updated the last paragraph with the information you mentioned, because I'm going to link to the post from the official Blizzard forums. Thanks!

[Quark]

It's a table, one roll. I can't completely remember the priorities, but here's a summary, +3 mob just to get the 40% glancing out of it:

Assume 5% Dodge/Parry/Block/Miss to make showing the numbers easier. Also assume 20% crit. I use 0-99 here, but it's more significant figures than that, as shown by the fact that you can have 23.45% crit.

000-004 -> Miss
005-009 -> Dodge
010-014 -> Parry
015-019 -> Block
020-059 -> Glancing (+weapon skill doesn't negate the % of glancing , it negates the damage penalty of a glancing)
060-079 -> Crit
080-099 -> Hit

+Hit is actually a misnomer. If a player gets a "+1% hit" item, it's truly "-1% miss". That's because a "hit" percentage doesn't actually exist. Your hit chance is simply whatever is left of the table after everything else is filled in. So the same table would become:

000-003 -> Miss
004-008 -> Dodge
009-013 -> Parry
014-018 -> Block
019-058 -> Glancing
059-078 -> Crit
079-099 -> Hit

+Crit consumes +Hit. So take the first table and add +1 crit, you get ...

000-004 -> Miss
005-009 -> Dodge
010-014 -> Parry
015-019 -> Block
020-059 -> Glancing
060-080 -> Crit
081-099 -> Hit

I believe crit is lower in priority than miss, dodge, parry, block, glancing. So in this case, adding 30 crit to the first table would give you:

000-004 -> Miss
005-009 -> Dodge
010-014 -> Parry
015-019 -> Block
020-059 -> Glancing
060-099 -> Crit

Your mitigation chance is the same, and now every hit is a crit. In this case, getting +1 hit will result in you still not hitting but having an additional crit, while getting +1 crit won't help you at all.

Late addition: okay, from looking over other places, it seems priority is:
Miss >> Dodge/Parry >> Block >> Glancing >> Critical >> Crushing >> Hit

Of course, glancing and crushing can never be in the same table since glancing is player attacking mob specific and crushing is mob attacking player specific.

[MongoJerry]

Nice tables. So, what would the tables look like if you had a debuff on you that reduces your chance to hit by 85% (anubisaths in AQ40) or by 75% (the first boss in AQ20)? Let's say the person has a 30% crit rate.

[Warlock]

Recklessness can give a Warrior 100% crit on white damage. Swings can still miss with it up. I've read that they can glance as well, but haven't tested it myself.

[Chesspiece_face]

Nice tables. So, what would the tables look like if you had a debuff on you that reduces your chance to hit by 85% (anubisaths in AQ40) or by 75% (the first boss in AQ20)? Let's say the person has a 30% crit rate.

I would think it would depend on what blizzard considers a "hit" for those buffs. using the tables that Quark listed and keeping the 20% crit rates the same etc. if these debuffs only effect the "hit" section of the table than it would look something like this:

000-021 -> Miss
022-026 -> Dodge
027-031 -> Parry
032-036 -> Block
037-076 -> Glancing
077-096 -> Crit
096-099 -> Hit

if "hit" refers to all attacks that land and do damage it would look more like this:

000-073 -> Miss
074-078 -> Dodge
079-083 -> Parry
084-087 -> Block
088-091 -> Glancing
092-095 -> Crit
096-099 -> Hit

these numbers a bit off because i'm just winging it, but you can see how the -%85 chance to hit might effect the tables. in the example above the normal chance to land a blow (glancing, crit, hit) would be 80/100. So 85% of those hits (68) would then become misses. without doing more research or seeing the exact code it's not possible to know how the remaining % chance to hit would effect Glancing, Crits, Hits etc. but even using the above example you can see that it wouldn't necissarily gaurantee crits on every hit.

[MongoJerry]

I would think it would depend on what blizzard considers a "hit" for those buffs. using the tables that Quark listed and keeping the 20% crit rates the same etc. if these debuffs only effect the "hit" section of the table than it would look something like this:

That seems unlikely, since it appears that really +hit% items (and therefore I assume -hit% debuffs) really act as -miss% (and I assume +miss% respectively) items. There's some discussion on my guild's boards and on the Elitist Jerk's forum that there might be some kind of miss% cap. I guess Wowwiki says that there's a miss% cap at 60% but no evidence is given to suggest how this was determined. (You can't believe everything on Wowwiki -- so much is wrong there).

Assuming that there is a 60% miss cap and that the target is high enough level where let's say there's a 10% chance of glancing blows, then I would guess that the table for someone who has 30% crit would look like this:

000-060 -> Miss
061-065 -> Dodge (assume 5%)
066-070 -> Parry (assume 5%)
071-080 -> Glancing
081-099 -> Crit

Glancing blows can't crit, so that would explain the non-crit damage. In this case, the person would have only a 20% chance to crit, because 10% of his or her theoretical crit was pushed off the table. That's my first stab at it, anyway.

[Quark]

Oh, that reminds me! Most combat log parsing addons don't distinguish Glancing from Hit. I modified my CombatStats so that it would (I only use CombatStats when I'm trying to gather some %s on damage, typically).

When seeing if there's hit or not, make sure it's not a glancing hit;)

[MongoJerry]

Oh, that reminds me! Most combat log parsing addons don't distinguish Glancing from Hit. I modified my CombatStats so that it would (I only use CombatStats when I'm trying to gather some %s on damage, typically).

When seeing if there's hit or not, make sure it's not a glancing hit;)

Does the combat log tell you if it's a glancing hit? Or would an addon have to do some kind of calculation to figure that out?

[Icebird]

Does the combat log tell you if it's a glancing hit? Or would an addon have to do some kind of calculation to figure that out?

For player hits against mobs, yes.

"Warrior hits Nefarian for 12 damage (glancing)"

Very easy to see when attacking a mob significantly higher in level than you.

Chris