[Elemental] ESSE stat spreadsheet (updated to 4.0.1)

47 posts in this topic

Posted

I was wondering how much chance Chain Lightning has to proc LO on 1st, 2nd and 3rd target.

This is a rough assumption (that's why it's red). The idea was that, before the LO buff, CL had the 20% proc chance only when it was hitting 3 targets. If it hit only one target, proc chance was around 7-8%. 8/20*33 = 13.2%, and with 13% chance the probability of not proccing for 3 targets is (1-0.13)^3 = 0.65, aligns well with the 33% overall proc.

As for the other reds, I had done no testing on them, that's why I left them as open questions.

I can't see any reason why overload procs of LB would have different scaling than normal LB casts

My understanding is that Lightning Overload procs makes the shaman cast another spell: Overload LB or Overload CL. These spells are indeed appearing in the parses if you use World of Logs, StatisCL, etc. Their base damage is the half of the normal LB and CL (only rounded), and as pre-shamanism tests showed, they have a SP coefficient of 35.7% and 28.6% - or how we calculated back then, you take the normal LB/CL, apply spellpower with the normal 71.4%/57.1% coefficient, and take the half of the damage.

Now Shamanism simply increase the coefficient of all Lightning Bolts, including these, that's how it became 45.7%, instead of half of the normal, Shamanism-enhanced LB's (71.4+10)%.

Share on other sites

Posted

This is a rough assumption (that's why it's red). The idea was that, before the LO buff, CL had the 20% proc chance only when it was hitting 3 targets. If it hit only one target, proc chance was around 7-8%. 8/20*33 = 13.2%, and with 13% chance the probability of not proccing for 3 targets is (1-0.13)^3 = 0.65, aligns well with the 33% overall proc.

I can confirm the approximately 13% proc chance on single target, as I've been doing extensive testing of this just after the proc chance was changed.

Share on other sites

Posted

Made a few more tweaks, here is the heads up:

• Added a trinkets tab. Hopefully added all Wrath caster DPS trinkets, let me know if you miss something.
• Changed Lightning Overload calculation with Clearcasting, as Mightyatom suggested. Didn't make a thorough testing, but I haven't seen any data contradicting this in any of my raid logs.
• Removed some "questions". (Thanks to Tufy and a guild mate of mine)

As always, any general comment is welcome.

Share on other sites

Posted

OK so here is my gear: The World of Warcraft Armory

I have used your spreadsheet a few times now and it seems to think i should be breaking 4k dps....I agree...but the only time i ever get that high is when all trinkets have popped and heroism is up. I just don't get these 5 - 6k dps elemental shamans out there. I have looked everywhere and can't seem to figure out how to get that high. There was an elemental shaman in my group the one day with the same gear (maybe one or two upgrades) and he was pulling 5.5k dps. I talked to him and i was doing the same things he was...same glyphs...same priority rotation ...everything. it was all single target dps on 25 man Patch and my computer can run WoW on ultra in any 25 man setting so its not lag / computer issues....i just don't get it.

One last thing, I just tried to dl the new version of this spreadsheet. I am getting =B3+QUANT in C2 and a bunch of <#>+QUANT formulas. Also Im getting a File Error: Data may have been lost. Ive tried exporting it 3 times. Im using Office 2007....could the latest excel be breaking it?

EDIT: Yeah so the latest Excel was sorta breaking it...it didnt like working with the tags....i fixed it now...and it says i should be doing about 5k dps....ha.....im obviously doing something wrong :\

Share on other sites

Posted

OK so here is my gear: The World of Warcraft Armory

I have used your spreadsheet a few times now and it seems to think i should be breaking 4k dps....I agree...but the only time i ever get that high is when all trinkets have popped and heroism is up. I just don't get these 5 - 6k dps elemental shamans out there. I have looked everywhere and can't seem to figure out how to get that high. There was an elemental shaman in my group the one day with the same gear (maybe one or two upgrades) and he was pulling 5.5k dps. I talked to him and i was doing the same things he was...same glyphs...same priority rotation ...everything. it was all single target dps on 25 man Patch and my computer can run WoW on ultra in any 25 man setting so its not lag / computer issues....i just don't get it.

One last thing, I just tried to dl the new version of this spreadsheet. I am getting =B3+QUANT in C2 and a bunch of <#>+QUANT formulas. Also Im getting a File Error: Data may have been lost. Ive tried exporting it 3 times. Im using Office 2007....could the latest excel be breaking it?

EDIT: Yeah so the latest Excel was sorta breaking it...it didnt like working with the tags....i fixed it now...and it says i should be doing about 5k dps....ha.....im obviously doing something wrong :\

Well to start with your gemming is horrible. You are well over the hit cap so those hit gems do nothing at all, you would get more benefit putting a green spellpower gem. Also you use MP5 gems which you should never ever seen on an elemental shaman as we shouldn't really ever go oom.

The real 'secret' between average and good ele dps is being able to use your fire DPS totems. Of course you should only do this is the buffs from totem of wrath are already covered, but always take advantage of your other totems if this is the case. The 3% crit from your totem does not stack with the same buff provided by a ret pally, and the spellpower is less than that provided by a demonology warlock. So if both of those are present, drop a searing or magma totem for additional personal dps, as your totem of wrath is doing nothing whatsoever.

So to boost your dps;

Use straight spellpower gems where possible, spellpower/haste gems where yellow socket bonus is worth getting, spellpower/stam gems if you need to satisfy meta requirements

Don't use cheap enchants - it won't make a huge difference but its a personal hate of mine seeing people using sub-optimal enchants. Grind exalted with sons of hodir for a better shoulder enchant, grab an epic leg patch for 15 more spellpower and enchant 10stats on your chestpiece.

Remember to only refresh flame shock just before a lava burst - you should get 2 lvb casts off in the duration of one flame shock, but just keep firing off lightning bolts and dont bother refreshing FS until a second before you can lvb again

Other than that I guess it just comes down to practice, good luck.

Share on other sites

Posted

Thanks a lot. Seriously it was insightful. So for yellow gems you would say haste > crit? and only if the socket bonus is worth keeping eh? Well I can agree with that. Do you bother throwing in chain lightning into your mix? Unfortunately A) my guild doesnt have any demo locks...only destro....B) no other ele shammies so no luck on getting another totem of wrath dropped. So i pretty much drop my ToW every time except for when its a group >= 4 mobs then i drop my magma.

So in recap ill change up my gems to maximize spellpower / and then see what i can do about upgrading my enchants ;).

Thanks for the help :)

Share on other sites

Posted

The link to the sheet is changed so update your bookmarks! On the old link you can still access the previous version.

- Added preliminary support for procs/on-use items: There are currently Bloodlust, Elemental Mastery, the new totem's proc, Orc/troll racial. Trinkets are still only modelled on their separate page.

- Added T9 and a bunch of trinkets.

- Small tweaks here and there.

Cooldowns are currently calculated like they won't overlap. This is not what you usually want to do (rather blow all cooldowns together), this is for easy calculation. (Any idea trying to model procs running together is welcome.) Due to this, don't worry if the no-cooldown part's "up-time" is negative; that means you cannot possibly avoid overlapping, not like this is what you want to do.

Share on other sites

Posted

We're sorry, [name]@gmail.com does not have permission to access this spreadsheet.

You are signed in as [name]@gmail.com, but that email address doesn't have permission to access this spreadsheet. (Sign in as a different user or request access to this document)

Is it currently locked?

Share on other sites

Posted

Meh. Note to self: Press apply before closing the browser tab, after setting up permissions. It should work now.

Share on other sites

Posted

Whenever I set "Is Troll" to True 2.75 is subtracted from the value of int, sp, haste, and crit which results in them all going negative.

Share on other sites

Posted

Fixed, thank you.

Share on other sites

Posted

I have finally updated the calculator to 3.3.

Just like at 3.2, the link has changed, here's the new one:

To support the new T10 set bonuses, I had to redo the "rotation" part of the sheets from scratch. This (plus the fact I had less time than I would like for this) sadly means that the current rotation does not include Chain Lightnings, for now. Also the support for troll and orc racial is gone for now. The complexity of this stuff suggest me to move on to a simulator, or - like for resto shamans - a combat log parser for future elemental shaman theorycraft...

Was a late feature of the 3.2 sheets too, although I did not announce it here, that there is a "hit helper" tab, where you can optimize your gear for loosing the least possible DPS stats to be hit capped. Usage is like: fill your candidate items' stats plus the few extra settings in the yellow fields, then play around with the blue fields - put true or false in them to select candidate item #2 for that slot - expected DPS is in the low-right corner.

Support for Flame Nova is missing, I need some theorycrafting on how that thing works for all possible amounts of targets, what is the SP coefficient, etc. etc.

I have made a Google Docs folder for the calculator, where you will always find the latest and all older versions, I suggest to update your links to this: https://docs.google.com/leaf?id=0B5clvbTTjpCDNzZjNWJkMDMtZDAzOC00NGNkLTg2NWYtOGVkYjI3YTMzOTIy&hl=hu (The link above is a direct link to the spreadsheet)

Share on other sites

Posted

Support for Flame Nova is missing, I need some theorycrafting on how that thing works for all possible amounts of targets, what is the SP coefficient, etc. etc.

Assuming you mean Fire Nova, the coefficient should be (and seems to be, though I've not properly tested it yet) 21.43%, and for damage boosts it should get 15% from call of flame and 20% from improved fire nova (the concussion tooltip doesn't say that it affects Fire Nova). It also procs, consumes and fully benefits from clearcasting (one single charge per cast, no matter how many targets are hit). And as it's AoE, the total damage caused per cast is capped at the equivalent of 10 targets.

Share on other sites

Posted

Assuming you mean Fire Nova, the coefficient should be

Yeah, ment that. Have you (or anyone) tested it's damage on single (and while at it, any number < and > than 10) targets? I somehow have the feeling it's not always 10x(tooltip+SP*coefficient), ive seen numbers that suggest it's even better for single target than LB, an other numbers that tell it's only useful in aoe situations, altho I also didn't have time to do a clean testing.

edit: typo. damn.

Share on other sites

Posted

I've only done a 5-minute test so far, naked in my resto spec using Flametongue Totem (144 spellpower). The expected damage results for a 21.43% coefficient would be

MIN: 144*0.2143+893 = 923

MAX: 144*0.2143+997 = 1028

The lowest damage figure I observed was 923 exactly and the highest 1023. So I am pretty confident about the coefficient.

I'm also pretty confident about the way the damage modifiers from CoF and IFN stack, as the tooltip with both reads "1205 to 1346 damage" - which is precisely the expected range for a 15% + 20% additive increase (893*1.35 = 1205.55, 997*1.35=1345.95).

This puts it below Frost Shock in terms of single target effectiveness. e.g. for 2500 spellpower and not accounting for crit (they have same crit chance/modifiers):

FrS av. base dmg: 835

Coefficient: 38.57%

Talents: Concussion (5%) and Booming Echoes (20%)

835 + SP*0.3857 * (1 + 0.05 + 0.2) = 2249

FN av. base dmg: 945

Coefficient: 21.43%

Talents: Call of Flame (15%) and Imp. Fire Nova (20%)

945 + SP*0.2143 * (1 + 0.15 + 0.2) = 1999

What I don't know is how the 10-target AoE cap works.

Share on other sites

Posted

What I don't know is how the 10-target AoE cap works.

Take spell's maximum damage on single target, multiply it by 10 - that's your maximum damage on any number of targets. On ten or fewer, you won't feel a difference, it's the eleventh mob that begins lowering the damage per target so that the total maximum damage remains 10 x [single target maximum damage]

Share on other sites

Posted

I have updated the spreadsheet to reflect the 3.3.3 PTR changes to Flame Shock. A few stuff is still needs testing and fix if necessary:

• 4T10 (currently always 2 extra tick)
• 2T9
• How does the new [iTEM]41531[/iTEM] stack with Elemental Fury

This version has also lots of tweaks and fixes here and there, and defaults were changed significantly, so results are not directly comparable to previous versions.

As a side note, there is also a 3.3.2 live version for weeks now. Updated the OP so people won't go to outdated versions.

Also question: What would you think about a tool like shaman_hep from stassart, i.e. a combat log parser, but for elemental (or actually any caster DPS in the long run)? I have got a veeeeeeeeeeeery early alpha version for that, would like to know if people would like to use such a tool.

Share on other sites

Posted

I think a personalized parser for our stats would be VERY helpful. I toyed with trying to put it together about a year ago and found myself just not skilled enough to do it. Provided its easy enough to use, it would go a long ways towards perfecting our gemming optimization and other little quirks that happen with subtle gear changes.

Share on other sites

Posted

Yeah, I agree with Moshne - I think a log parser would be a fantastic tool if you can get it to maturity, and very interesting to compare with the sims and formulations. Could potentially help bridge the gap between theory and praxis.

Share on other sites

Posted

It lives!

I have updated ESSE to 4.0.1 - details are in the updated original post. However, there is one issue I already know of.

I am not really sure about the new spell power coefficients - I have found this spreadsheet, but not really know if that contains the info I need - is it Live or PTR/Beta? lvl 80/85? Does it include stuff like Concussion/Elemental Precision?

Anyone know / tested the new coefficients?

Share on other sites

Posted

This might be what you're looking for. These are the raw datamined coefficient values used by simcraft.

Share on other sites

Posted

This might be what you're looking for. These are the raw datamined coefficient values used by simcraft.

Just to clarify:

sc_scale_data.inc - simulationcraft - Project Hosting on Google Code contains things like combat ratings, intellect to crit scalars, and the class and level based multipliers that are used to determine things like the base damage and damage range of most spells.

If you're after the actual spell *coefficients* then you need to instead look in:

That file lists all of the spells and spell effects that SimulationCraft is currently importing from the DBC files (which is pretty much every class ability and as many pet abilities we can find).

The format of the data in sc_spell_data.inc is detailed in:

data_definitions.hh for those familiar with C/C++.

SimulationCraft of course has access functions to simplify getting things like the coefficients etc. for a spell but to give an example of how it can be used, let's find the coefficient for Lightning Bolt.

Here's how we can go about it.

1) Find the spell id for Lightning Bolt. As there can be a few abilities and effects with the same name it's important to make sure we're starting with the right one. So first, let's check sc_spell_lists.inc. This gives a list of all the "clickable" abilities for each class giving the spell id. We see the line:

```
403, // Lightning Bolt

```
i.e. The spell id for Lightning Bolt is 403. 2) Search the spell list in sc_spell_data.inc for spell 403. The line is:
```
{ "Lightning Bolt"                    ,   403, 0x00, 20.0, 0x08,  0, 0x040, 0x000,  7,   1,     0.0,    30.0,       0, 1500,    0,         0,   6, 0x000,   0,   0,   0,  0,  1500,  2500, 20, {   151,     0,     0 }, { 0x00010000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000 }, "Casts a bolt of lightning at the target for \$s1 Nature damage.", 0 },

```
3) This a pretty simple spell. It only has 1 effect and we can see from \$s1 that its damage is based off that 1 effect. The effect id for effect 1 is 151. We also see that the Scaling Type is 7 (which is the Shaman one). We now find effect 151 in the effect list. It has the line:
```
{   151, 0x00,   403, 0, E_SCHOOL_DAMAGE             , A_NONE                                     ,  0.7670000196,  0.1330000013,  0.0000000000,  0.7139999866,     0,     0.0,     0.0,      12,       0,       0,     0,   0.0, 0.400,  3 },

```

Now, the relevant numbers in this line for us at present are: 0.7670000196, 0.1330000013 and 0.7139999866.

The first is used to calculate the average base damage and is also used to calculate the base damage range.

The second number is also used to calculate the base damage range.

The third number is its coefficient.

4) Next we need to check sc_scale_data.inc to get the correct multiplier for our level. Let's say we're level 80.

We know that Lightning Bolt is using Scaling Type 7 (i.e. Shaman) so we check __gt_spell_scaling[][] for that entry and get: 900.120300292968750.

5) We can now calculate the average base damage, its damage range and of course we have the coefficient already.

The average base damage is the first number in 3) multiplied by the number from 4).

i.e. 0.7670000196 * 900.120300292968750 == 690.3922879670649169921875.

The damage range is the first number in 3) multiplied by the second number in 3) multiplied by the number from 4).

i.e. 0.7670000196 * 0.1330000013 * 900.120300292968750 == 91.822175197129608317145329589844

Thus the minimum value is: avg - range/2 and the max is avg + range/2.

i.e. min == 690.3922879670649169921875 - 91.822175197129608317145329589844/2 == 644.48120036850011283361483520508 which on the tool-tip iirc would be 644.

max == 690.3922879670649169921875 + 91.822175197129608317145329589844/2 == 736.30337556562972115076016479492

which on the tool-tip will likely be 737 (But it might only be 736, I forget. Doesn't matter too much as the actual amount will vary between the floating point numbers).

So for a Level 80 Shaman, Lightning Bolt has a base damage of around 644 -> 736 and has a spell power coefficient of 0.714 (pretty much all of the coefficients are set at 3 decimal places and it's just the inaccuracies of floating point numbers that makes them slightly different).

Not every spell is as easy as something like Lightning Bolt of course, but it's an example of how the data values are used by the client. It's important to remember of course that all of these figures come from the *client's* DBC data. It's the information the client uses to determine what numbers to show on the tool-tips. The server of course can have completely different values with hot-fixes and the like but usually this isn't the case. And if a spell doesn't show any damage ranges in its tool-tip at all (say with the Flame Elemental's melee attacks I guess) then usually that also means there's no data in the client DBC files for it and the numbers for that need to be tested manually and can't be just pulled from the DBC files and quickly verified.

If something is still unclear about my example let me know.