You don't need to be in a raid to show critical hits are being suppressed. As long as you know you current critical strike chance, anything will work. Out of raid, my critical strike chance with Imp Scorch, right now, is 45%. That should be plenty high. And then I could test without Imp Scorch and see the difference.
"gear" is a particular set of equipment. I am keeping the same equipment in each slot for most of the tests, and trying to match Crit exactly and hit approximately between Inco and Exactly.
"Level" is my level.
"TrgLvl" is target level, boss assumed to be 83.
This data taken primarily from Recount Screenshots, which I have available if needed. I also have combat logs of most of it. The labour intensity of parsing so many so large files has forced me to prefer the Recount approach.
I have run tests with Exactly as I have been leveling her. She happens to have a 15 spell penetration headpiece, I also tested the effect of Spell Pen on Crit suppression at level 80 vs Boss level. Doesn't seem to be any, or very small.
The clearest trend is in the 4 tests 4,5,6,7. This shows that for that particular gear (198 Crit rating, 155 hit) on a level 80 target, Crit suppression is 1.21% per level difference converging to level 80 where there seems to be no Crit suppression on a level 80 target.
Tests 8,9,10,11 are on a Boss level target, same gear. Fitting a linear on this data shows a crit suppression of 0.76% per target level difference. However this is somewhat dubious. The "noise" is very high. I am sceptical that it even is noise, the behaviour on Boss level might be very non-linear
Test 15, 16 (suffering from a low population sample) show a totally different level dependency again, and possibly non-linearity.
What we need is something that ties in: level difference, base crit chance, and crit suppression in the one equation, and isolates the exceptions and non-linearities.
Not very complicated, just a flat 2.5% crit suppression for most people, less if your Crit rate is under 10% ish.
I shall continue the level dependency testing with Exactly, and get some more data points with crit rates 2-15% with Incoherent. I'll also test the effects of Molten armour and other buffs. There is obviously more to it than this, caster level and so on.
I have been following the thread and I understand the statistics involved. I also know that a targeting dummy is not a boss. Even GC has said he would rather see actual fight numbers over target dummy numbers for any performance based accusation. If anyone has a group of friends that are dedicated to theorycrafting, take a group in and do AM's for 6 mins until he enrages and do this 2 or 3 times.
Any mage can get 200+ Arcane Missile attacks in a 3 min fight. If 10 people do this twice we have a nice sample set of 4000 data points. This will not give us conclusive data, but it will give us something to compare to the data we already have to see if there is any consistency.
It might be worth trying the same sort of thing on Loatheb. He's got a much longer enrage timer, and the spores could allow for the possibility of reaching 100% crit, which should show diffinitive proof that there's a suppression if any missles don't crit.
It might be worth trying the same sort of thing on Loatheb. He's got a much longer enrage timer, and the spores could allow for the possibility of reaching 100% crit, which should show diffinitive proof that there's a suppression if any missles don't crit.
From PTR bugs in the past (Stat stacking = fun), we know the paper-doll crit rate can exceed 100%, so that would be entirely dependent on when the reduction is applied, and if the calculation is maxed at 1.0 crit rate.
Incoherent, you know that table of yours would make for a great first post in a thread about this, since it definitely applies to all specs, not just FFB...
To truly model the game, we first must research it. http://zaldinar.wordpress.com/
Proven TheoryCrafting Stuff, chain casting in a PTR near you soon.
Incoherent, you know that table of yours would make for a great first post in a thread about this, since it definitely applies to all specs, not just FFB...
It doesn't quite feel complete yet. I'll do that though when I get a few more data points. Exactly is accumulating Rested XP at the moment for the next AoE level grind.
One interesting observation. I did a test last night where I took off all Crit gear, leaving Intellect as the only contribution to Crit chance. Unbuffed, I had a Paper Doll Crit Chance of 5.07%. 3000 or so casts showed no Crit suppression. On the contrary, Crit rate was inflated. Would like to explore this a bit more.
It doesn't quite feel complete yet. I'll do that though when I get a few more data points. Exactly is accumulating Rested XP at the moment for the next AoE level grind.
One interesting observation. I did a test last night where I took off all Crit gear, leaving Intellect as the only contribution to Crit chance. Unbuffed, I had a Paper Doll Crit Chance of 5.07%. 3000 or so casts showed no Crit suppression. On the contrary, Crit rate was inflated. Would like to explore this a bit more.
By inflated, how much was it inflated by? Was it inflated by a significant amount?
Good work, btw and thanks for following up. I've been a bit busy recently to dedicate tie to practice and several of my runs have been ruined by paladins and fire mages. I have recently respecced and will be exploring what difference 15% flat crit (5% from Molten Armor, 10% from Imp. Scorch debuff) has on crit suppression. Hopefully I can do the test sometime this week...
I'll have to wait until I get home from work to verify, but from memory it was in the region of 0.5% - it's likely to be an anomally, or a passing Firemage, must check to see if I /combatlog'ed it. However, it may hint at the possibility that "natural" crit chance, i.e. that given by Intellect (and perhaps buffs) is not suppressed, indicating that the suppression is to Crit Rating rather than Crit Chance. Perhaps, "your Crit Rating on targets above your level is reduced by 30 x level difference to a minimum of 0" kind of thing.
The inflation was only 0.35%, 5.42%, within expectations for the crit chance of 5.07%. It does suggest that there is no suppression in this case. It was 3960 casts, Crit rating 0, no buffs so only Crit from Intellect.
I am by no means competent when it comes to developing addons, but if someone here is, an addon that tracks all hits/crits/current crit percentage will help a great deal with this situation.
It could be similar to WWS but much simpler and specifically aimed at Mages and tracking crit rating. Many of us here on the forums would be willing to run it to get larger sample sets. This might be a tall order or it might not, I simply don't know or understand the details of such an addon. It would need to track your casts' hits, crits, misses, and any +crit buff or debuff. At the end of a night of raiding, you would have a decent sized set of data to analyze.
THAT would be incredibly useful. Tracking Crit Chance, Hit chance, Intellect, Spellpower, Hit rating, Crit rating per cast ... would be worrying close to cheating.
On a whim at work I checked that idea of the crit suppression being on Crit Rating rather than overall chance. I'll be goddamned if suddenly the data doesn't all fit. With a 36 odd crit rating per level difference reduction, and a 10-15x divisor for targets below caster level, all the data I have where I have a number for Crit rating, fits a linear, expected vs experienced, with an R2 value of 0.974...
Would like to see some more numbers. Especially with buffs and with Crit Ratings crossing around 3x36ish, 100. My scattered constellation of data points that I have so far, for all levels I've tested can be predicted with a sum squared error of 5.6% from 16 data points.
I am not an Addon Dev. but i am a programmer, if we open a new thread with a list of things this new addon should do, i'll have a look into some addons code and try to put down something useful.
I guess having Recount as a base sould help alot but we need a list of features, if possible sorted by priority, to look at; i don't want to just rename Recount if it can do all the things we need, i want a list of things that other addons are missing and are usefull for us
Thanks if someone want to contribute, sorry for my english, i am ita
I would suggest using something really lightweight for a codebase...so that players can have those other recording methids running, and still collect information too.
I dabbled in addon developing a little while ago. While I do not have the time to commit for this addon specifically, I can confidently say that anyone who does won't have much trouble. All you have to do is register for the relevant spell events (missed, crit, hit), and keep track of them.
The main reason is not only that we found a systematic crit reduction, but we also set out to establish sample set procedures, and appropriate sample sizes in order to reduce the amount of mathematical deviation within a sample set. We found this eliminated a lot of the uncertainties with the testing.
We removed outliers, and did derivations on the results. We achieved results within ~0.2 percent fluctuation from expected results.
The main aspect of the test is that we ran 12,000 shot tests for each variable, in sets of 1000. We then looked at the variation in returns, and determined that even a sample set of 5000 events could fluctuate as much as 2 percent, causing most returns within that threshold to be deemed inconclusive due to mathematical circumstances.
Agreed it would not be hard to track. But we also have to look at the variations between spell and physical crit ratios and what affects them.
To my knowledge (On an average basis), naxx bosses have no spell resistance, so that means that the diminished spell hits (The random 2700 FB over the standard 4k), will be playing a factor in something. Whether it be the actual crit number, or if it is in fact reducing our chance to crit. Perhaps it's an intended suppression? Perhaps they do have resistances. Doing a 5k cast set, I noticed about 30 partial resists, why are they there? What is their purpose? To simply reduce our dps? Or is it something the target has we're not accounting for.
We can produce the addon to track all of these. And i'm sure someone somewhere would be more than happy to sit there and cast fireballs all night to produce a sample set. But the real question someone has to answer, is what they're coming from.
We've all noticed that the reduced crit returns are happening, question is, is it a bug?
I'd like to do some testing on the PTR, should anyone care to join me, and test this change to Spirit -> Crit. The way I see it, they're changing it away from a flate rate gain, to an inflated ratio. Perhaps that is what's doing it? The Flat 5% is somehow being suppressed by the fact that the bosses level is simply higher than ours and changing it to a Conversion Gain will alleviate that problem?
To my knowledge (On an average basis), naxx bosses have no spell resistance, so that means that the diminished spell hits (The random 2700 FB over the standard 4k), will be playing a factor in something. Whether it be the actual crit number, or if it is in fact reducing our chance to crit. Perhaps it's an intended suppression? Perhaps they do have resistances. Doing a 5k cast set, I noticed about 30 partial resists, why are they there? What is their purpose? To simply reduce our dps? Or is it something the target has we're not accounting for.
All mobs higher level than you have a slight resistance chance (2% per level if I recall correctly) that is not mitigated by spell penetration. This is why you see partial resists. It is comparable to melee glancing blows.
Edit: Wording.
Originally Posted by Crowl
If you have to control a robot dinosaur that fires lazers and there's a time when you shouldn't be shooting those lazers then the encounter is clearly flawed beyond hope of fixing.
Of course. But the question is, could the partial resistances have a crit reduction effect as well that we've simply overlooked in the past?
Spell penetration won't help this partial resistance, because it's not an actual number to a school of magic, it's simply a level difference. Could the same be said for crit reduction effects?
Edit: Will transfer this discussion to the proper thread.
First of all, let me appologize if I'm asking you to repeat what's been said, but I'm trying to help a fellow raider not myself, and the idea of reading 11 pages of stuff on mages when I play a warlock seems like a major waste of my time.
I have looked her up and she appears to be a FFB build. 0/52/19. I do have a mage (frost, 78, non-raider), so I'm pretty sure the fire talents look decent, she may have a few too many in frost, but the mana reduction from those 3 in the 4th tier can be useful.
She's typically somewhere around 14th in damage and newer (ie less geared) mages are outdoing her and she hasn't really improved over time. One thing I've noticed is that she's not hit capped. On a recent patchwork fight, she had a miss rate of around 7%. I know for a warlock, this would be intollerable, but I'm not 100% sure for mages. I do remember that deep fire used to be able to sacrifice hit for crit. Is this still the case?
On this same Patchwork fight, her damage was spread out like this:
FFB 46% (52% crit, 6% miss)
Ignite 21%
Pyro 17% (79% crit, 7.1% miss)
Living Bomb 9% (40% crit)
Remaining on scorch, lightweave and mirror images.
She actually came in 9th for that fight (about 3300 dps), but for the evening, she was 14th in overall damage and 12th in overall dps (2697) for all bosses. At that point, I believe she had 4/5 valorous.
I haven't really seen that much improvement in her performance. Is this expecting too much out of this build? Should she be worrying more about hit than crit (I armoried her, she has 3/3 in the hit talent, but is only sporting 8.5% from gear). What about haste and spell power?
Again, I appologize for my ramblings and my non-specific questions, but I really hope you can help me help her.
First of all, <Blah blah blah blah blah> I really hope you can help me help her.
Ok, firstly, make sure she has 11% hit assuming you are raiding with a ShadowPriest or Boomkin. If not, make it 14%. That should be the first priority.
The gear otherwise, after gaining hit cap, is: SP > Crit (Only just) > Haste.
Since we Frostfire mages use Burnout, crit is essential but SP still scales better.
Secondly, make sure the gear is gemmed and enchanted correctly. Red sockets use +19SP, Yellow sockets take +9SP +8 Crit, Blue sockets take +9SP +8Spirit. If she is a jewelcrafter, make sure she uses all 3 of the Runed Dragon's Eye at any one time and make sure they are used in Blue Sockets > Yellow Sockets > Red Sockets (Highly unlikely).
All gear should be enchanted too:
Head slot from Kirin'Tor
Shoulders from Sons of Hodir (Or inscription)
Cape with Haste or Lightweave Embroidery
Chest with +8/10 stats
Bracers with +23 SP (+67 with LW)
Gloves with +28 SP OR Precision (Increasing hit if required)
Belt with an extra socket (+19SP)
Boots with Icewalker (If you need hit) or Minor Run Speed (If you are moving more than 4 seconds per minute in any given battle)
Rings with +19SP each if enchanting.
Weapon with +50SP or +63SP if BiS or can afford it.
Lastly, make sure the rotation is correct. This is crucial. Her opening should be:
LB > Scorch > Scorch -->
but she will not have a rotation, but she should priorotise her spells in this order:
LB > Instant Pyro > Frostfire Bolt > Scorch.
I can't think of anything else required to maximise DPS, other than get Quartz for the latency bar at the end. I suggest sending her to this post (#1073) on this forum to read it herself, there is quite a lot of information here.
Being under hit cap isn't really tolerable for any class, although completely gimping yourself simply to hit cap isn't advised (but shouldn't be an issue anyways, the gear is easy to come by).
It's really hard to say without being able to armory this person or see a WWS. I imagine they aren't keeping LB up as much as they should if it's their fourth damage dealer, mine is typically third after Ignite.
edit: this came after Yttiriums much more informative post. Though, you should check which glyphs she is using. Should be Frostfire, Molten Armor, and Scorch.
I took the liberty of armorying your guild's mages. As you haven't volunteered a name, I'll assume you want it to remain private, and I'll respect that. I've identified a few changes for the one mage in your guild who *fits* the profile you've provided.
2) She is way under the hit cap. Replace [Potent Monarch Topaz] with [Veiled Monarch Topaz] until she reaches the hit cap, which is likely going to be 264, assuming there is a draenae in the group. When the gear improves, she can gradually regem with potents.
3) Drop Herbalism!!! If she is serious about doing decent dps, this is a must. Leave gathering to an alt.
4) Steal a point from Improved Blizzard and max out Piercing Ice.
5) Lastly, be patient...FFB spec is very gear dependant, and she is no where near complete. I can see she is stacking SP and Crit, but sadly at the expense of Spellhit. Her priorities should be:
SP>>Spellhit(to cap)>>Crit>>Haste
Until she replaces a lot of her placeholder pieces (MP5??) with better gear, she will be doing substandard dps.
04/05/09, 1:26 PM
