Edit 2: Looking back at the stat conversions in previous versions, I feel like an utter noob. Haste is valued ridiculously high in those older versions, and I'm suprised I could not find an issue or discussion on it already. Moving on...

In 2.2.0b3, it looks as if haste is being applied to the cooldown of Lava Burst. Either that, the tooltip over "dps" is wrong, or I just severely fail at simple math.

With my current gear, the dps tooltip is stating 7.8 Lava Bursts per minute. That's a Lava Burst every 7.7 (or so) seconds! Even a haste capped shaman should have a minimum of 9 seconds between LvB's, right? 8 sec cd + minimum of 1 sec cast time?

Also, here's what 2.2.0.3's stat conversion selection is telling me:

hit: 4.22
sp: 2.59
haste: 2.46
crit: 1.29

Those numbers are in my current gear (you can see for yourself to the left of course) with every raid buff/debuff except for focus magic and draenai racial. Obviously the hit is there b/c I'm at 286. Buffed stats:
2528 spell power
286 hit rating
53.19% crit (on LB/CL)
533 haste rating

I was under the impression haste was not as close to spell power in relative value. Is the oddity in the dps tooltip connected with the relative stat values shown?

Wodi, how would you suggest I implement the Improved Shield talent?

Incoming damage is not modelled, so orbs don't proc.

JoeHeff, the current (beta3) version has a misleading tooltip. Instead of "casts per minute" what you see is "seconds per minute spent casting". If you divide it by cast time, you get a correct number. I'm correcting this, so in the future you will actually see "casts per minute".

Does that mean that haste is being correctly valued so closely behind spell power? I'm not so good with the complex math, so my basis for gear has been to use bink's custom weight scale for wowhead. I've been playing with rawr as a cross reference, but TBH I don't know how to apply the math to verify if there is indeed a descrepency btwn the stat weights of these 2 sources.

Thanks for all the time and effort you put into this module! (same goes for bink and the many other contributers here at EJ. I'd be lost without it! )

Basically, calculations go as follows.

1. FS - Lava Burst - n Lightning Bolt - Lava Burst - m Lightning bolt
2. Lava Burst - FS - n Lightning Bolt

Rotation 1 if you have Glyph of Flame Shock, rotation 2 if you don't.

n and m depend on haste. Now, n and m can be rounded up and down. Obviously, you can't cast half a spell. Most common values I see are n = 4-6, m = 3-5. If rounded down, this means there will be short moments of doing nothing, if rounded up, this means lava burst and flame shock will not be cast at maximum efficiency.

Haste is valued a lot less now.

Besides, you will see that on the SEIC, the DEP for haste does not remain static, but rather varies with your current haste.

That's kind of the trick with these things - you'd like to see things more dynamically, e.g.

1. if all other gear is held constant, what are upgrades in each gear slot
2. what is the best-in-slot total combo to shoot for
3. if x, y, and/or z best-in-slot combo pieces are unavailable, what is the next-best total combo to shoot for

Rawr 2.2.0b4: Selecting Glyph of Flame Shock causes relative item values to change drastically, example to follow:

Character used for testing:
Rounded on Stormreaver-EU
Selected full raid buffs, Flask of Frost Wyrm, Haste food, Glyph of Lava, Glyph of Lightning Bolt
Fight duration: 5:0
80% of fight spent in replenishment
Gem setups:
- Runed/Runed/Runed/Runed/Chaotic (rare)
- Veiled/Rigid/Royal/Rigid/Chaotic (rare)
- Runed/Runed/Runed/Runed/Chaotic (JC quality)

Deathchill Cloak with 23 Haste enchant has a value of 107,78 / 323,35
Pennant Cloak with 32 Spellpower JC gem and 23 Haste enchant has a value of 120,57 / 361,70

Selecting Glyph of Flame Shock causes value to change to these with same gems and enchants:
Deathchill Cloak 231,14 / 693,43
Pennant Cloak 85,59 / 253,75

What causes these values to change so drastically?

I will attach the profile file in the next few minutes so manual entry isn't needed.

[e] http://arakki.fortunecity.com/simcraft/ROUNDED.zip for Rawr 2.2.0b4

You're right on the typo, I'll fix it.

Regarding the trinkets: here's what happens:
1. We calculate, without procs, the expected number of casts/hits/crits/misses/dot ticks.
2. Using this information, the trinket proc parameters (I manually programmed these) and the fight duration, the expected uptime in seconds and the expected number of procs is calculated. Using the expected uptime as a % of the total fight length, the average value of each trinket is calculated.
3. This is used to recalculate the expected number of casts/hits/crits/misses/dot ticks.
4. This procedure is repeated, because haste procs will result in more casts which results in higher uptime which results in a higher average haste rating which results in possibly even more casts.

Parameters used in calculation
[Mark of the War Prisoner] duration 20 seconds, iCD 120 seconds
[Dying Curse] duration 10 seconds, iCD 45 seconds, 15% on cast
[Sundial of the Exiled] duration 10 seconds, iCD 45 seconds, 10% on hit
[Embrace of the Spider] duration 10 seconds, iCD 45 seconds, 10% on cast

You can see that [Sundial of the Exiled] and [Embrace of the Spider] are expected to have similar uptime.... (around 1/6) [Embrace of the Spider] adds 84,17 haste rating first pass, 84,41 haste rating second pass.... [Sundial of the Exiled] adds 98,33 spell power first and second pass. I used mostly best in slot items for this calculation.
This gives:
[Sundial of the Exiled]: 84 crit rating and 98 spell power
[Embrace of the Spider]: 84 haste rating and 98 spell power
Now, isn't that interesting?

The Relative Stat Values display in Rawr shows that crit > haste, so apparently Sundial of the Exiled is better. At least, according to the model. So unless you can show me where the model is bad, it appears the sundial is a better item than embrace of the spider. Again, show me if the model is bad, that's what I'm concerned about mostly.

And as a bonus: why do I think Rawr can get away with averaging the values? Because we're stuck with a pretty static rotation that's basically controlled by cooldowns on spells: FS/LvB/nLB/LvB/mLB repeat. If a trinket procs, you're not suddenly going to cast different spells to make maximum use of the proc.

I can't download your profile file.

What you're actually looking at are "compared" values, which are indirectly calculated. They represent the value of the item in certain conditions, but if these other conditions mean you can do twice the DPS, your items are suddenly worth twice as much too. When comparing the relative values of individual items things get even more interesting.

In your case, I suspect it has to do with the rotation. When you select Flame Shock, Rawr will calculate a FS/LvB/nLB/LvB/mLB ad nauseam rotation. If you don't have it, it will go for LvB/FS/LB ad nauseam. What I see from the Relative Stat Values with your character, is that Haste Rating suddenly gets very interesting without the Glyph.

You have a bug in the calculation of the Lavabust crit damage from what I can tell. Just downloaded rawr from the site and loaded my profile, by adding the 4pc bonus the damage on LvB is something like 53000 for crits. My dps essentially doubled from adding 1 item, just swap the benefactor's gloves for the heroes' earthshatter gloves and you will see it

Quite a bit so. After optimization, including [Sundial of the Exiled], I show crit being worth almost four times what haste is valued at. This is the opposite of b3, where it actually gemmed some yellow sockets with sp/haste gems.

I wish I could be more helpful in determining that. The latest SEIC (3.1.0.10) with a Priority+ rotation shows haste worth almost twice the value of crit (1.013>0.609). I don't know where or how the calculations could differ that much.

Is that the Beta 4 version or 2.1.9? The "official" release is not the current build that's being tested.

Trolando, it looks like what you have created is now suffering a bit from what I like to call robot santa syndrome (if you've seen The Santa Clause 2...)

Before, haste seemed overvalued a little, but it is now being potentially hugely overvalued or hugely undervalued. In all fairness, it doesn't seem to be an error, but rather a product of the somewhat locked down rotation (especially since CL isn't in the module at all). Anyway, when you hit the 5/4 LB sweet spot at just a bit over 500, items start getting a bit out of control.

For example, no cloak has more haste on it than the Deathchill Cloak, right? And the 2 ilvl 200 rings from heroics are pretty stacked with haste as well. Well if you're just barely over the sweet spot (you can import my setup from armory... Apirin/Bronzebeard as shown in my profile), it can make what I would think are upgrades look like garbage in comparison. This seems to be a result of the new option to round n and m up or down. Looks more to me like "rounding down" is actually truncating. (Well to be fair, reading the option makes it sound like that anyway...) Because of this, if I swap say Deathchill cloak with Pennant Cloak and Annhylde's Ring with Signet of Manifest Pain, I lose a good 300 DPS. (The simulation section shows that I'm down to 4/4 instead of 5/4).

Is there a way that you can give it some sort of breathing room? What would be cool is if there is a slider bar type option that the user could use to specify how long they are willing to "wait" for LvB cd's or overlap a LvB cd. Would that be more complex than adding an option to throw Chain Lightning into the mix? One or the other (or both!) would be a great addition to help smooth over the RSS (robot santa syndrome!) that the module is applying to haste now.

Thanks!!!

EDIT:
What you are seeing is along the same lines of what I'm seeing. I believe it's a result of the LB's being "rounded" up or down. Essentially without enough haste, Rawr is telling you to wait 1.5 seconds for LvB cd because your would take 1.7 seconds to cast (just pulling sample numbers for illustration's sake). That is if you uncheck the "round up" options. If you check those options, you'll basically see the opposite happen... that is, if your LvB cd would be up in, say .1 sec, it's going to tell you to cast that 1.6 (or however long) LB instead of waiting.

Binkenstein and I seem to disagree on how the rotation should be calculated. We both work with a FS/2LvB/nLB rotation, but while he works with a real number n, I work with an integer number n (resulting in FS/LvB/nLB/LvB/mLB because I then need to split n into two integer numbers n and m). The theory is that he suggests that a rotation will have n lightning bolts on average, while I believe that a player will either always choose to cast an extra bolt, or always choose to wait a bit to cast lava burst as soon as it's off cooldown, thus resulting in an integer values of n (e.g. 8, 9, 10, while Binkenstein allows for values like 8.435). Perhaps you can compare Binkenstein's sheet with a fixed number of lightning bolts in the rotation and see what you get from that. (My calculations aren't 'fixed', they scale with haste, but they are rounded up or down depending on the user preferences) Also, Binkenstein uses "casting delay due to lag", while I haven't implemented this option yet. If you really want to compare Binkenstein and Rawr, set this delay to 0 ms and use a fixed number of lightning bolts, equal to n+m in Rawr.

I was about to throw CL in the mix, then I heard that in 3.1 CL will be a lot less interesting because it's single target dps will be nerfed; it would be more efficient to cast an extra lightning bolt and let LvB wait a bit. I would think that this 'robot syndrome' may just be something we have to live with: haste is still interesting, especially if you round up, because more haste means the rotation takes a bit less time per cycle. It is a bit like sets in this way: if you already have one tier item equipped, the other tier items are suddenly a lot more interesting.

Anyway, even if I modify Binkenstein's spreadsheet to round values up, it shows different numbers. Maybe someone else can go take a look at the module too? It's not that complicated.

You can import the character profile too, then again my gear got changed a bit tonight so it'll be off in terms of showing the possible problem.

Would it be possible to add an option to use LvB without the glyph so FlS would get consumed? Or should we trust the values we get with the FlS glyph?

T-

What I was trying to suggest was instead of always rounding up or down, have a give or take option. Ie I don't care if it's up or down as long as I'm within .5 or so seconds. (in practice I get my rotation more precise because I have a static haste value that I am experienced with, but upgrading gear changes that haste value so it makes the module clunky for comparing possible upgrades)

That's already there. Just select some other glyph besides GoFS. It just doesn't reflect the change in the rotation text.

Will using this method trap you into a set rotation? From what I can tell with my poor math skills is haste will always be devalued over crit. Meaning the module will always weight stats based on a set number of LBs. Until it is forced, through higher item budget (226 and beyond), into a higher haste range adding another LB to the rotation, the gap between haste and crit will get larger.

I know it has been discussed earlier that Rawr helps in gear selection, not DPS simulation. If that is the case, would it be better for it to use real numbers vs integer numbers? The way it calculates now is an absolute perfect (read Patchwerk) situation. A step in any direction and your rotation is off.

On a side note: Is there a way to add the option to use real or integer numbers? That way people could have the option based on their view of the best method.

Yes. For me rounding up is a 718 DPS increase. I wish.

I have a question. If MP5 is not that great for an elemental shaman why is it rated so high in rawr? This is really confusing me when trying to select gear with rawr.

Use common sense Nakawe, are you running out of mana? If not, you probably don't need the suggested MP5. Maybe also read this one more time. The section on Other Stats was quite enlightening for me.

Did you check the fight duration, and make sure you have a replenishment buff enabled? Also make sure RAWR imported your spec properly. If it did not, you will have issues with mana resutling in MP5 being rated much higher than the rest of the stats.

I had fight length set for 10 minutes. I guess what I forgot to take into acct was that I am not casting for the whole 10 minutes, if i was I would need more mp5. I was basing it on 10 minutes because Uldar fight are about that. Thanks for helping me clear it up though.

Trolando, where you be?

Tro hasn't posted in quite a while. The last update on this module seems to have been a over a month ago. I was curious if anybody else had any interest in picking this up? I wish I could. Rawr is getting better and better with each release, but the module is falling behind.

Just curious. It seems like a nice tool in addition to the spreadsheets we have available.