Mists of Pandaria: All Specs - Page 4

Ruga · 04/15/12, 7:31 AM

I noticed a few things on my Beta Enhancement Shaman (Build 15589):

- Fire Elemental also spreads Searing Flames debuff on the target it attacks
- Enhancement AOE is pretty crippled as it currently is. My shaman has 20'000 mana - that's 4 Novas if I don't "hardcast" something.

Titanian · 04/16/12, 7:28 PM

Fire Nova is almost certainly bugged, or else Mental Quickness has an incorrect tooltip. MQ is supposed to reduce the cost of "your instant beneficial, damaging, and totem spells" by 75%. Fire Nova's current cost in Enhancement spec is 5140. When you switch to Elemental or Resto and page back to the inactive Enhancement spell list, however, Fire Nova is still listed at 5140, instead of the 20560 it would be if it was previously under the effect of Mental Quickness. (To see this for yourself, switch from Enhancement to Resto or Elemental and check Earth Shock's and Feral Spirit's mana costs before and after.) This means that FN was never under the effect of MQ in the first place, despite meeting all the criteria for it.

Since there's no real indication that they're changing Mental Quickness to exclude Fire Nova, I'm leaning towards 'bug.' Make sure to submit a bug report.

Argannor · 04/17/12, 10:02 PM

Originally Posted by Cheysuli

I calculated rough proc chance from Lightning Bolt (80 casts) and Earth Shock (50 casts). It seems proc chance is about 36-38 % (very rough due to small sample).
here

I did some testing by myself to confirm the high proc chance of this talent. I casted 100 LBs, which gave me 8 procs (8% chance therefore). While testing some other mechanics I felt like this 8% should be ok; 37% is way to high.

I also tested how Echo does interact with other mechanics (about 10 min. per mechanic).
I wasn't able to gain a proc, which got duplicated by the Mastery: Elemental Overload (maybe due to the low proc chance: about 4% at my mastery level). Even though I was able to gain LS stacks from Echo procs. DTR duplicates didn't proc the Echo (or vice versa).

Tested on Beta Build: 15589

Cheysuli · 04/18/12, 3:25 AM

Originally Posted by Argannor

I did some testing by myself to confirm the high proc chance of this talent. I casted 100 LBs, which gave me 8 procs (8% chance therefore). While testing some other mechanics I felt like this 8% should be ok; 37% is way to high.

I also tested how Echo does interact with other mechanics (about 10 min. per mechanic).
I wasn't able to gain a proc, which got duplicated by the Mastery: Elemental Overload (maybe due to the low proc chance: about 4% at my mastery level). Even though I was able to gain LS stacks from Echo procs. DTR duplicates didn't proc the Echo (or vice versa).

Tested on Beta Build: 15589

I believe Blizzard changed proc chance in new build (15589). I was trying to test Echo on monday and proc chance was much lower. Secondly I believe (yet I have no proof) that proc chance depends on your specialization. In enhancement spec I get much more Lightning Bolt procs than in restoration spec. Unfortunately I can't provide any logs, because mekkatorque realm and beta client is very unstable and I got several disconnects and critical errors during testing.

Argannor · 04/18/12, 8:28 AM

Originally Posted by Cheysuli

Secondly I believe (yet I have no proof) that proc chance depends on your specialization. In enhancement spec I get much more Lightning Bolt procs than in restoration spec.

Well, I tested the proc chance in restoration spec, because it's much easier to see the procs when there are no aditional Elemental Overload procs.

Cheysuli · 04/22/12, 7:56 AM

I finally did more testing of Echo of Elements. Wow Beta build is 15589. I used lvl 85 tauren Shaman. Testing specs were enhancement and restoration. That is important, because proc chance of this talent depends on your specialization. Damage spells were tested on lvl 85 Training Dummies (with one exception) and healing spells were tested mainly on myself and in one case on random afk hunter (lvl85)

I was wondering whether blizzard changed if certain spells can proc this talent and I found that Unleash Frost, Lightning Shield (static shock), Searing Bolt, Flametongue weapon proc and Frostbrand weapon proc can't be duplicated in enhancement spec. In restoration spec I couldn't duplicate Lava Burst, Unleash Flame and Unleash Life (didn't tried others), but that could by issue of very low proc chance. I don't know if its intentional or not, nevertheless it will be reported on the beta bug forum.

Now to the data. To determine proc chance I was using mainly Lightning Bolt (no cd, cheap, no dot), but proc chances of other spells were tested as well.

Enhancement spec:

Lightning Bolt: 76 procs out of 300 casts (25% proc chance)

Lava Burst: 24 procs out of 75 casts (32% proc chance) - lava burst now has 100% crit chance even without flame shock on target in enha spec (normal behaviour in resto spec). Great fun insta killing mobs in Twilight Highlands. Reported as a bug.

Chain Lightning: I tested this spell on one, two and three targets and results are at least interesting.
1 target: 9 procs out of 30 casts (30%)
2 targets: 22 procs (2 hits) out of 30 casts (73%) - note: only 2 available training dummies are lvl 80 dummies in Thunder Bluff.
3 targets: 17 procs (3 hits) out of 45 casts (38%) and 10 procs (2 hits only) out of 45 casts (22%) - 60% proc chance combined

Different proc chance on 2 targets and three targets might be result of small test sample, but it could indicate that proc chance depends on level of target. I need to do more tests on this.

Healing Surge: 6 procs out of 25 (24%)
Chain Heal: 9 procs out of 25 (36%) - only one torget.

Proc frequency of other tested spells was roughly same.

Restoration spec:

Lightning Bolt: 9 procs out of 100 (9%)

Greater Healing Wave: 7 procs out of 100 (7%) - partly tested on random afk hunter

Healing Surge: 9 procs out of 100 (9%)

There is significant difference in proc chance between two specializations. Enhancement echo of elements duplicates damage and healing spells roughly three times frequently than in restoration spec. I can't determine proc chance in elemental spec, because respecing is not implemented yet and elemental mastery also adds another dimension to this problem.

If you are interested in the raw data, you can find them here.

pfooti · 04/24/12, 12:58 PM

A brief word on reporting proc rates. Simple procs (no ICD, no scaling of rate with haste via a PPM mechanic) are estimated by regular binomial variables (as an example, a fair coin is a binary variable with a p=0.5).

If you're reporting observed proc rates ( $\hat p$ ) of X procs out of N observations, the observed proc rate is obviously (X / N). That part is easy. There's a lot of talk about how confident you are about those values (how big of an N do you need?)

If you're really interested, there's a decent writeup here.

The short story is, to have a 95% confidence that the actual proc rate is somewhere within your reported range of proc rates, you report $\hat p \pm z_{1- \alpha /2} \sqrt{ \frac{\hat p \left ( 1- \hat p \right )}{n}}$ . In our case, $z_{1- \alpha /2}$ for a 95% confidence is 1.96. So, for example, 76 procs out of 300 casts has an n of 300, a $\hat p$ of 0.253. Therefore the 95% range of confidence is $0.253 \pm 1.96 \sqrt { \frac {0.253 * 0.747}{300} }$ . Therefore the proc rate is between 0.278 and 0.228. Probably 0.25, since programmers like multiples of five.

If you'd prefer a 99% confidence interval (meaning you can say that it's 99% likely that the ACTUAL proc rate is within your reported values), change the 1.96 to 2.58.

Naturally, if you can get 300+ observations, this isn't a big deal. But it helps get a handle on corner cases like when you're testing changing proc rates under some buff or another and it's difficult to get an N above 20.

Cheysuli · 04/24/12, 2:34 PM

Thanks for that math pfooti. I take my data as preview of this talent so we have an idea how good it might be and how does it work. To get at least a bit accurate proc rate I need 1000+ casts as a sample.

Anyway I did another test. Whole testing was done in enhancement spec. I was interested if proc chance depends on target level. I used Lightning Bolt against lvl 85 training dummy and lvl 60 training dummy. I got 66 procs out of 200 casts in first case and 65 procs out of 200 casts in latter. I think that this theory can be rejected.

I also did more tests with Chain Lightning.
Against 1 target (lvl 85) I got 34 procs out of 100.
Against 2 targets (lvl 80) I got 48 procs out of 100.
Against 3 targets (lvl 85) I got 59 procs out of 100.

This finally shows that proc chance is increasing with number of targets (might be interesting to test with Glyph of Chain Lightning). I also found what those strange numbers of duplicated hits mean. When testing against 3+ targets I sometimes get +2 hits and sometimes +3. Whole problem was caused by position of training dummies in Orgrimmar. There are 4 dummies in row. I was targeting first one and sometimes proc hit 3rd one jumped to 4th and wasn't able to jump back to hit 2nd dummy.

pfooti · 04/25/12, 2:25 AM

Well, if the proc rate is actually 0.35 and you have a additional 0.35 chance of proccing a dupe off of a proc, you'd expect to see an overall proc total of (0.35 + 0.35 ^ 2) * 100 = 47, so your two-jump number seems reasonable along those lines. Adding triple procs, you'd expect to see 60, so all those numbers seem to make sense. You're still getting an overall proc rate of 0.35, but the way the system counts (some double and triple procs), it makes the proc rate seem higher.

Note that those are total procs, meaning that you'd still only get 35% of your chain lightning casts to proc events, of which 35% are doubles and the remaining 65% are singles.

Cheysuli · 04/25/12, 3:27 AM

Originally Posted by pfooti

Well, if the proc rate is actually 0.35 and you have a additional 0.35 chance of proccing a dupe off of a proc, you'd expect to see an overall proc total of (0.35 + 0.35 ^ 2) * 100 = 47, so your two-jump number seems reasonable along those lines. Adding triple procs, you'd expect to see 60, so all those numbers seem to make sense. You're still getting an overall proc rate of 0.35, but the way the system counts (some double and triple procs), it makes the proc rate seem higher.

Note that those are total procs, meaning that you'd still only get 35% of your chain lightning casts to proc events, of which 35% are doubles and the remaining 65% are singles.

Actually no. Number of procs I wrote is number of casts where duplication occured. Number of hits is always maximum possible. That means when attacking three targets with 100 Chain Lightnings I got 300 regular hits plus 177 (59*3) additional hits. As far as I know Echo of Elements duplicates whole spell.

• Jessamy · 04/25/12, 7:18 AM

If CL Echo were to provide the same 35% chance to proc the whole spell per target, the expected proc chance would look like this:

targets chance for no Echo on cast chance for Echo proc
1 0.65 35.0%
2 0.65^2 57.8%
3 0.65^3 72.5%
4 0.65^4 82.1%

This is not the behavior you're observing, so if targets beyond the first provide an additional chance to proc the whole spell, it's a lower chance.

Alternatively, the game doesn't have to follow simple binomial probability rules; stochastic observables follow the behavior of a game system, not the mathematical rules of a (virtual) real world physics. For example, the game could detect server-side how many targets are hit by the original cast, then have a table it checks to see what the chance of a proc is with that many valid proc targets, and finally, roll against that table. A lookup table solution is less elegant than a formulaic solution, but we don't know the right model to represent the empirical data yet.

Ruga · 04/25/12, 12:50 PM

They have "fixed" enhancements mana gain for now it seems.
Never got below 29k (max of 32k mana on my 87 shaman). They also added a lot of new subtle lightning effects to Stormstrike.

Also for the Echo discussion that's going on: I haven't confirmed it myself, but I observed that it also seems (Visually at least) to affect Fire Nova resulting in a boatload of AOE. I noticed foremost that there's always an explosion after all the others have gone through.

Cheysuli · 04/25/12, 1:16 PM

Originally Posted by Ruga

Also for the Echo discussion that's going on: I haven't confirmed it myself, but I observed that it also seems (Visually at least) to affect Fire Nova resulting in a boatload of AOE. I noticed foremost that there's always an explosion after all the others have gone through.

That is strange as Echo of Elements tooltip clearly says that it affects direct damage spells only. I'll try to confirm it and report if it's true.

Cheysuli · 04/25/12, 5:51 PM

I tried to duplicate Fire Nova on two adds in Jade Forest and it definitely works. When having two targets, one with flame shock, Fire Nova did damage to one target and again after a second. I managed to duplicate that spell many times and with flame shock on all targets. Reported.

Pyrdon · 05/12/12, 10:34 AM

I looked through all the available glyphs and found Glyph of Flame Shock interesting to analyse for a dmg bonus. (/edit tooltip is wrong - just click it for the correct tooltip)

So glyphed FS dmg will be 0.75*[initial dmg] + 1.25*[periodic dmg].
Each tick will do the same dmg with or without the glyph.

This results in:

Each one application of glyphed FS does more dmg than unglyphed.
Over the same time window, unglyphed FS does more DPS than glyphed FS (assuming same uptime), because each apllication does more dmg and we may have more of them.

Regarding Enhancement:

Using the glyph gives us 1 free shock CD every 4 FS (= 120 seconds = 5 unglyphed FS) - thus giving us 1 more ES - while losing 4 times on the difference of the initial dmg and 1 complete initial dmg.
So, what's better?

Some numbers from the beta:
My FS does something about 3.4k Initial dmg and about 11k periodic dmg unglyphed.
Glyphed numbers change to 2550 initial and 13750 periodic.
Initial dmg drops by 850.
ES is at almost 7k.

If I manage to get an extra ES every 4 FS I gain I'll gain:

Optimal case: 7000-4*850-3400 = 200 dmg over 120 seconds = 1.666DPS (Woooh yeah!).

Anything else is a DPS loss (uhhh =( )

I admit, that's just nothing. Adding UE_Flames to this increases the absolute difference even more und might turn the whole into a complete loss.
But something to add:

Glyphed FS (30 seconds) fits better with our UE CD (15 seconds). Maybe this is be interesting for someone.
The 30 seconds will also be spread via LL. (Tested on SW dummies)

Regarding Elemental:
Well, since Ele does not use shocks on CD basis regulary, we need another model:

We can compare the freed GCD to LB as a filler.
Its safe to asume that every 2 free GCDs (240 secs) will give you enough time to cast another LB. (under optimal circumstances with enough haste or proccs you could gain 1 LB every 120 sec)
Using the numbers from a premade Elemental shaman:
Loosing 2600 - 2000 = 600 FS initial dmg on each apllication.
LB dmg: 12k

So for 2 free GCDs: 12000 - 8*600 - 2*2600 = 2000 over 240 seconds = 8.333 DPS
For 1 free GCD: 12000 - 4*600 - 2600 = 7000 over 120 seconds = 58.333 DPS

As a little extra bonus: your management of Fulmination might get easier too.

Facit:
Blizz is doing a good job at keeping overall dmg nearly inaffected by glpyhs or talents.
Anyway it is nice to know that Enhancement will most of the time lose with the FS glyph while Elemental has a minor gain.