refuse

Regarding elemental precision

I did some empirical testing over the weekend and think the data is pretty conclusive
Elemental precision is bugged and it gives Frostbolt 3% extra hit (but not Fireball).

Data~
All tests were done with a level 65 character shooting rank 1 spells at Dr Boom with zero hit rating from gear.
(recount screen shots can be provided in case anyone cares to check for clerical errors, although i awesomely forgot to get one of test 3)

1. frostbolt - no talents
hits 3948
crits 195
resists 869
miss 17.3%

1b. frostbolt - 35 assorted filler frost talents
hits 2040
crits 65
resists 442
miss 17.4%

2. frostbolt - +5 empowered
hits 2096
crits 202
reists 496
miss 17.8%

3. frostbolt - +5 empowered +3 precision
hits 15915
crits 1495
resists 2192
miss 11.2%

3b. frostbolt - +5 empowered +3 precision +5 winters chill
hits 3725
crits 940
resists 544
miss 10.4%

4. fireball - no talent points
hits 4249
crits 136
resists 890
miss 16.9%

5. fireball - +3 precision
hits 4924
crits 184
resists 833
miss 14%

I think this pretty much speaks for itself.


While looking at my data i found two other things worth mentioning.
During tests 2, 3 and 3b i recorded the immune to frostbite messages and found that the proc rate was above 15% in each of the tests by ~1.5% each time.
Combining the 3 tests for frostbite relevant data gives...

total casts 27605
total connects (hit + crit) 24373
frostbite procs 4015
procs/connects 16.5%

I dont know anything about what is a reasonable amount of variation in this size of sample, but it seemed pretty odd.


Also perhaps even more interesting than the elemental precision results were the crit results.
Unfortunately i didnt control for crit, so the data is not very good.

In every test my observed crit rating was 0.5-2% smaller than expected.
In the tests where i had talent points providing +crit the observed was ~90% of the expected number, and in the tests where i had no +crit talents ~75% the expected.

This suggests to me that either my sample was far too small, dr boom was reducing my crits by a set value, or the character tooltip is faulty.

For anyone who may want to look at the data themselves...

1.
hits 3948
crits 195
6.57% character screen crit
0% from talents
4.71% observed

1b.
hits 2040
crits 65
4.25% character screen crit
0% from talents
3.09% observed

2.
hits 2096
crits 202
4.25% character screen crit
5% from talents
8.79% observed

3.
hits 15915
crits 1495
4.25% character screen crit
5% from talents
8.59% observed

4.
hits 4924
crits 136
4.25% character screen crit
0% from talents
3.6% observed

5.
hits 4248
crits 184
4.25% character screen crit
0% from talents
3.1% observed

*3b.
hits 3725
crits 940
6.57% character screen crit
15% from talents
20.15% observed

*This set involved some winters chill ramp ups and the character screen crit possibly dropped as low as 5.37% due to gear breakage. I believe that there were only 2 winters chill ramp ups and that the crit loss due to gear breakage affected 0-5% of the casts, however that is only my educated guess and this set is undeniably tainted~

 

Muphrid

Refuse: great work. A couple things though...

1) Without a test for Ice Lance, we may never know if this is a Frost-wide bug, or merely a Frostbolt bug. Since Fire has no binary nukes, we will definitely never know if this is a Frostbolt bug or a binary nuke bug, should Ice Lance get no benefit, unless someone's masochistic enough to cast some Frost-based AoE.

2) The crit data does suggest crit decay against higher level mobs, as has been suspected. I don't think your samples were too small--the differences are large compared to the standard deviation.

 

ReignConfused

I'm not sure who figured the exact amount of haste rating to make it scale better. But as long as we are arguing "infinite gear level" Pyroblast is the single best scaling spell in the game.


Edit: I found it in Vontre's Sig (I knew I saw it before!)

<manly>: pyroblast spam is viable with 3140 passive haste rating

So as long as we are arguing infinite gear level, yes Pyroblast scales better than every other mage spell in the game.

 

◊ manly

that was more of a joke than anything else. 3140 spell haste will make your pyroblast 1.5s cast time.

recommended change Mirror Image causes the caster to generate 75% less threat for the duration. (rather than a temporary -4m threat)
<blizzard> Your threat is divided among the images while [mirror image] is active.
<blizzard> Mirror Image divides your threat among the images. I thought this was well known
<blizzard> Okay, I look like a jerk. Mirror Image doesn't work this way.

bug [Dark Rune][Demonic Rune][Fel Mana Potion] works on vezax.
PSA WWS does not calculate most pets.

 

Muphrid

Joke or not, that's more indicative of the flaw in spell haste's design than any intrinsic advantage of Pyroblast.

 

Zalbo

Blastwave is binary i believe, not that it's at all practical to test that.
Blizzard isn't binary, and hits every second, so you could test that even quicker than ice lance (assuming AOEs are effected the same)

 

Muphrid

Good catch! Blast Wave would be a nightmare to test, though, yeah. Flamestrike, perhaps?

Blizzard wouldn't be too bad. Indeed, for Refuse to use the same toon, he'd have little choice: no Ice Lance at level 65.

 

Pintofbrew

Refuse:

Fantastic statistical analysis. At last someone had the patience to do hours upon hours of accurately presented data. Kudos and excellence for proving conclusively that we're getting 3% extra hit, and that it's from Ele Pre and not from any other baloney source we thought it was.

 

Leialyn

Humn, so they just "forgot" to "nerf" it for frost when it was reduced from 6% hit to 3% hit or is this something with binary spells...
Yep Muphrid, we could really need Ice Lance tests

 

Fizzl

Refuse, was this tested on PTR?

Im wondering blizzard going to stealth fix this without tell us or if they think this is how it should work.

 

Guaicow

Can anyone point me a thread that shows how much haste and crit is worth compared to +dmg?
Iv'e seen ppl say that 1crit rating is worth 0.65dmg, is this right? How is haste compared to +dmg and crit?
Sorry about bringing such a OT matter, i've used the search function with no luck and math was never my strenght.
Thanks in advance

 

Pintofbrew

Fizzle: Blizz hasn't fixed it since before BC when elemental precision became 3% rather than 6% (because: mage dmg = zomg as we all know) what makes you think they'll fix it now? Other than the possibility some blizz CM is reading EJ forums that is.

guaicow: the relative value of hit, crit, haste and damage constantly varies depending on:

1) How much of each one you have (including consumables)
2) What spell/rotation you're using
3) What spec you are
4) What debuffs are on the boss

There is no quick and easy response. It is quite complex. Rule of thumb is that crit is generally worth 0.7 damage, which is reasonably accurate in any spec. Barring the new CSD meta gem, crit has never been close to equal with +dmg.

Haste is a little more complex as it depends more heavily on the spell you're using than the other factors. I have found it is generally safe to assume 1 haste is between 0.8 and 1.1 dmg, depending.

Irrespective of anything of course, it is always optimal to stack +Hit till you're capped.

Remember that haste will not affect DPM at all, but it will affect MPS (mana per second) consumption, giving haste the added negative side effect of costing you in terms of resource, something which no other buff is affected by (appart from crit rating when specced elemental mastery, but even then the MPS and DPM change for 1Crit is negligible).

If you don't want to go whole-hog and start playing with Vontre's spreadsheet, then perhaps Lhivera's TC will be more helpful; given spec, conditions and gear it'll output the item equivalences for a given cycle which should help give you a ball-park figure:
Lhivera's Theorycraft Script

 

Guaicow

Very educational, thanks a bunch.
You brought up an intresting aspect of crit tho: the new meta.
I wonder how is crit going to be valued compared to dmg when using this meta, anyone run the numbers yet?

 

Fizzl

That's quite interesting. Being able to drop 39 points of hit and get 48 points of damage (assuming 4 +10 hit gems were switched with +12 damage gems) that means i'd be doing slightly more DPS as raid frost than as raid fire based on by Lhivera's Theorycraft-o-Matic (love this site) and my own gear.

Thanks EJ forums :D

Now back to choosing which bit of hardware to get that will spam frostbolt 1000 times a second ;)

 

Sancus

Of course, it's absurdly impossible to actually make a gear combination that will do this. Nobody is gemming for hit at the T6 gear level.

<Vontre> I removed the cooldown on evo
<sancus> and what happened?
<Vontre> DPS went down rofl

 

Muphrid

Lhivera's tool is great, but for pencil-and-paper theorycrafting, the equivalence between +damage, +crit, and +haste can be expressed with a formula.

 

Myrdinn

Frost is looking interesting but a lot depend on WE uptime.
Without pet, Vontre sheet give me "only" 1100 dps with my gear, and 1300 dps with fire.

We are only starting Hyjal/BT, but friends told me there is no way a WE could stay alive in Hyjal/BT fights...

Is that true ? or simply a "lack" of pet skills ?

 

Cardynal

Your pet could die on every boss except for Kaz'rogal (unless the pet can blow up from running out of mana) in Hyjal. Rage and archimonde aren't quite as hard on the pet.

I can only say up to RoS for BT. You should be ok for every fight except for Teron.

That's really the number 1 issue with Frost right now. Your water elemental needs to stay alive to do competive damage.

 

Fizzl

Lack of pet skills i think, though i've only done 4 in hyjal, 2 in BT so far.
I pop the pet just after the boss has done his big effect and it will usually survive its entire duration. You also need to be ready to tell him to follow you since hes flimsey and will need to move just like everyone else if there is an effect on the ground.

Ice block was very nice to have on the first BT boss

 

Belvedere

As long as we're on the topic of spell hit rating, could someone confirm my math on desirable +spell hit? I think I've got it right, but none of the analyses I've read explicitly address the Draenei racial. Am I right in thinking that my goal for +spell hit would be 113, if I'm a Draenei and deep frost?

(16% - 3% for precision, -1% for Draenei racial, -3% for precision+Frostbolt = 9% required. 9*12.6=~113.)

 

Pintofbrew

1) Don't get a G15 just to do the spam cast thing, it has been proven that it will cause a GCD to come up. Blizz has found this out and are introducing this to combat hardware-solutions like G15.

2) Nobody ever equips 10hit

3) If you're frost, untill Blizz comes up with "Elemental Precision: now this is WAI and won't give 6% to frostbolt" you don't need 164hit, you only need circa 126.

4) Don't trust Lhivera to guarantee your pet will look lively. Look once at the "frostbolt+waterbolt averaged" and three times at "frostbolt"value. Then do the same and reduce both numbers by 10% for spell pushback. Then think: Worst case scenario for fire, will be perhaps 1200dps. worst case scenario for forst will be 850dps. and decide if you can live with frost.

Note: Numbers were plucked at almost random. But an interrupted, pet-dead frost is indeed a dire sight to behold.

 

kadgar

I copied the numbers to a excel sheet to check the difference between 2 and 1 roll system for the crits but some numbers seem a bit strange:
-Number of fireball crits in tests 4 and 5 seem to be switched between top and bottom of post.
-I don't get your number of 27605 casts for frostbite test, wich values are you adding here?
-If you say 3948 hits, 195 crits and 869 resits you mean a total number of 5012 (hits+crits+resists) casts and 4143 (hits+crits) spells landed right?

 

Fizzl

OK back to pressing in the red zone

I have 1!(+10 hit gem) but it would still work 8 hit vs 9 damage gems.

Im frost because i like being frost :P

Frost bolt + WE average '1,174.13'
Frost bolt spam '1,047.22'

I can live with that, its not as if my WE dies alot, i've been playing with it since it was release so would consider myself to be an advanced WE user

.. what is spell push back?

 

koetjeka

Spell pushback occurs when you cast a spell and get hit by some enemy. Every hit you receive will increase your casting time by 0,2 seconds or so, when looking at your castbar you will notice a "pushback".

 

Qbert

Great work, Refuse!

Now the question for me that pops up is; Would it be plausible that frost and binary spells in general are getting double the benefit of hit% from all sources from a coding glitch with binary spells rather than just pointing the finger at elemental precision?

It may be unlikely but it is worth the effort to know for sure that it is undoubtedly the talent giving the bonus hit% and not a binary-spell bug.