Cataclysm Balance Theorycraft and Spreadsheet Thread - Page 6

klüger · 09/12/10, 7:58 AM

Thorns test at 85 with lvl 80 mobs in hyjal (about 200secs of thorn-impacts, no crits observed):

With 4388 sp (balance gear balance talents)
casterform, it hit for 2984.
casterform, it hit for 3371 with E&M
moonkin, it hit for 3413.
moonkin, it hit for 3857 with E&M

With 163 sp (feral gear, balance talents)
casterform, it hit for 671
moonkin, it hit for 767

With 155 sp (feral talents and feral gear)
casterform, it hit for 513

Erdluf · 09/12/10, 9:31 AM

Another Lunar stretch strategy would be MF spam (with LS). I think changing nothing else in your spreadsheet, considering only the stretching (energy neutral) portion of the rotation:

Wr+SF (49% SF by cast time) 4533 DPS.. Burns 5480 MP5
MF-spam (3*LS) 4328 DPS. Burn 127 MP5

Starfox · 09/12/10, 12:05 PM

Originally Posted by Arawethion

Yeah, the reason I was focusing on base values is that we got the coefficients from earlier in this thread. The data that someone just posted on TMR does confirm all the coefficients though (and it appears Starfall has been buffed back up somewhat). I got base values from his data; they match the Wowhead data to within a few points, so I'll put the Wowhead numbers into the sheet next time I upload it. They didn't change all that much really--I had figured there would be big changes from 80 to 85, the way everything else in the game increases so much.

I do want to look into Thorns more. With that coefficient, it's going to be nearing 4000 in decent L85 gear. That would seem to give it a very high DPET if you cast it someone who's tanking even one ordinary mob. If he takes only a few hits, the DPET is far higher than W or SF. I'm rather confident that it will have to be toned down somewhat. Regardless, it's an important spell to get tabulated in the sheet, since it will surely be worth casting a lot whenever there are multiple mobs around even if it's nerfed. And right now don't even know what buffs affect it, actually (can Thorns even crit? I don't remember).

My 85 druid thorns hit for 3100, only motw+MK, and i'm still in mainly ICC gear, think 2 dungeon blues, but on the other hand thorns is also one of the most expensive spells in my spellbook

You can also extract scaling numbers from the DBC files, it is what we do in the cataclysm branch of simulationcraft.
We have someone who extracts all the things we need into a datastructure we can just import and we gate basedmg, scaling, casttime etc. w/o doing anything. Takes one commit to update all spells, only changed mechanics need attention.

• Hamlet · 09/12/10, 12:54 PM

Yeah, I did that some during alpha when I wasn't in. I just never actually learned how to extract the data file from the client. At one point someone gave it to me though and it was kind of a fun project figuring what column represented what. If you and/or wowhead can get good values for spells reliably though, then maybe I shouldn't worry so much about experimenting heavily, just enough to confirm things (especially for complex spells like Starfall, which can be tricky to figure out from the spell data).

Originally Posted by Vaag256

About thorns, I just tried it in a duel on my level 85, while in feral spec;
With 3703 spellpower it hits for 1348.
With 151 spellpower (gearless) it hits for 343.

And versus a few level 9 raptors;
With 3703 spellpower it hit for 2006
With 151 spellpower it hit for 511

Hmm, why are these values different?

e: Well regardless, the bottom values match the tooltip damage/coefficient.

Erdluf · 09/12/10, 2:40 PM

In a duel you're up against things like Resilience, Perseverance, and Barkskin (or their analogues from other classes). Not really a good environment for theorycrafting PvE numbers.

Erdluf · 09/12/10, 5:54 PM

Originally Posted by Starfox

You can also extract scaling numbers from the DBC files, it is what we do in the cataclysm branch of simulationcraft.

I'll let simulationcraft do the hard work. Extracting from simc's cata branch sc_spell_data.inc (updated at simc site 9 hours ago), and filtering for Druid spells that have an effect with a coefficient other than one or zero, I get:

Deep Healing, 2.5
Entangling Roots, 0.1
Entangling Roots, 0.1
Healing Touch, 0.806
Hurricane, 0.095
Insect Swarm, 0.26
Lifebloom, 0.0234
Lifebloom, 0.355
Moonfire, 0.18
Moonfire, 0.18
Nourish, 0.355
Razor Claws, 2.5
Regrowth, 0.037
Regrowth, 0.367
Rejuvenation, 0.134
Savage Defender, 4
Starfall, 0.127
Starfall, 0.247
Starsurge, 1.535
Swiftmend, 0.536
Total Eclipse, 1.5
Tranquility, 0.068
Tranquility, 0.398
Typhoon, 0.097
Wild Growth, 0.042
Wild Mushroom, 0.286
Wrath, 0.714

For heals, I wouldn't be surprised if those are +heal coefficients (in LK, you'd multiply by 1.88 to get a +SP number)

Edit:
It looks like the Wild Mushroom coefficient is three times the Hurricane coefficient. In two GCD's, Hurricane ticks three times, or you could place a WM and detonate it. However in 4 GCD's you can place and detonate 3 mushrooms (or get six Hurricane ticks).

If those numbers stick, WM will be a better AoE, unless you specifically want the Hurricane debuff (lowered attack speed), or if it turns out that Hurricane has a larger radius.

Edit2: I forgot about Gale Winds. Direwolf at TMR says WM doesn't crit. The difference in AoE DPS potential should be pretty small right now.

• Hamlet · 09/12/10, 6:37 PM

Great--those line up perfectly with what's been seen in-game (except that Typhoon was apparently buffed by 30% since then*).

The only really mysterious spell remaining in FoN. Have the SimC guys looked into that at all yet? Some data I got from TMR shows individual hits scaling slightly more strongly than they do now, so we probably have to check whether the AP conversion is still the same and all that.

*Edit: I want to make sure the didn't accidentally have Gale Winds when he tested, but I think I've had the 0.126 coefficient in the sheet since back when Lissanna tested at level 80.

• Hamlet · 09/12/10, 7:33 PM

Ok, more sheet stuff (wow, I've been doing a lot of this this weekend).
--Popped in all the spell values that we've been discussing. Only spell that really needs info is FoN, and to a lesser extent Thorns and WM (what buffs affect these two?).
--Swapped priority of IS and MF since IS has much better DPET now. In reality I think good players with wind up fudging their cast timings slightly based on whether an Eclipse is coming, but it's still good to know.
--Removed some stuff that no longer exists (Moonkin form mana proc, JoW). Looking forward to excising all the junk in there that's related to NG, but trying to wait a bit longer, just in case.

At this point I'd say the spreadsheet, much like the expansion, is basically in beta. With all the important data and machinery in, you can now use it to model actual Cataclysm DPS and get meaningful results. Pending any further refinements, bug fixes, and whatever changes are necessary due to remaining beta patches, it is essentially ready for live and we can use to start disseminating information about optimal rotations.

klüger · 09/13/10, 12:45 AM

The wild mushroom thing is too early to tell about (I reckon), since it currently hits for 200-300dmg pr shroom @ 4400 spellpower. So most likely the scaling is simply not implemented.

Erdluf · 09/13/10, 8:51 AM

In the 912 spreadsheet it looks to me as if Starsurge (and Thorns, FoN, and Mushrooms) are ignoring Hit.

Starsurge ignoring hit was enough to make haste>hit at some gear levels.

I have no idea whose hit Thorns would use, and I don't know how (or if) our hit affects treants.

For those of us used to LK damage ratio's, putting this table (and a pie chart for extra credit ;-) on the rotations page is instructive:

Rotation Split	Damage	Time
Non-eclipse Wrath	8.0%	14.9%
Eclipse Wrath	17.1%	22.4%
Non-eclipse Starfire	4.5%	9.8%
Eclipse Starfire	17.0%	25.9%
Starsurge	11.8%	10.0%
Moonfire	8.6%	4.9%
Insect Swarm	21.8%	9.4%
Starfall	6.5%	2.1%
Force of Nature	4.6%	0.7%

Total 100.0% 100.0%

If I got that right, Insect Swarm did more damage than Starfire.

Glyphs: I'm seeing a number of situations where the MF glyph > SF glyph in the spreadsheet (keeping IS and Wrath as my other two primes). That is somewhat counter-intuitive, but it may be correct.

Edit: Followed Starfox's hint to make columns line up. Thanks.

Starfox · 09/13/10, 10:40 AM

Originally Posted by Erdluf

Rotation Split Damage Time

Non-eclipse Wrath 8.0% 14.9%

Eclipse Wrath 17.1% 22.4%

Non-eclipse Starfire 4.5% 9.8%

Eclipse Starfire 17.0% 25.9%

Starsurge 11.8% 10.0%

Moonfire 8.6% 4.9%

Insect Swarm 21.8% 9.4%

Starfall 6.5% 2.1%

Force of Nature 4.6% 0.7%

Total 100.0% 100.0%

Rotation Split	Damage	Time
Non-eclipse Wrath	8.0%	14.9%
Eclipse Wrath	17.1%	22.4%
Non-eclipse Starfire	4.5%	9.8%
Eclipse Starfire	17.0%	25.9%
Starsurge	11.8%	10.0%
Moonfire	8.6%	4.9%
Insect Swarm	21.8%	9.4%
Starfall	6.5%	2.1%
Force of Nature	4.6%	0.7%

Total 100.0% 100.0%
use | to use colums in the table markup, makes it way more readable.

• Hamlet · 09/14/10, 12:15 AM

Another update. Probably the last one for a few days, but it's pretty spiffy now. As far as substantive stuff, added Thorns and Wild Mushroom in (but still need better info on what buffs affect these, as well as on Force of Nature). A bunch of minor changes as well.

Erdluf · 09/15/10, 11:59 AM

I think the spreadsheet under-values Stretch rotations, and has some bigger problems with one-sided (Stretch+Unused) rotations.

The spreadsheet has

SolarUptime = (Time Spent Casting Wrath During Solar Eclipse) / (Total Time Spent Casting either SF or Wrath)

It is used to determine what fraction of OnCooldown spells are cast under Solar Eclipse (important for IS and SS).

That formula (and that use) is pretty good for a Standard rotation. For a Stretch solar, SF is also cast under Solar (to help stretch it). I think the equation would be more accurate as:

SolarUptime = (Time spent casting Wrath or SF during Solar Eclipse) / (Total Time spent casting either nuke).

In terms of the current spreadsheet the numbers are available as

Time spent casting either nuke under Solar = Time spent casting Wrath under solar + Added PreSolarTime

Similar changes would need to be made for Lunar Stretch.

For the one-sided Rotations (Stretch Solar + Unused Lunar) there are at most two pre-solar casts (actually these are between-solar casts). For (Unused Solar+Stretch Lunar) it takes at most three full Wrath casts to proc Lunar again. I'd be inclined to use one cast for preSolar, or two casts for preLunar for these two cases.

Something not modeled in the Spreadsheet (not easy to model either) is that Stretch Solar has a nice synergy with the Starfire glyph. If you are using MF on cooldown, and not using Stretch Solar, a fair number of MF casts would not be extended by the glyph.

All of this is still in the 200 DPS (out of 12000 total DPS). Not huge overall, but it can change which rotation is optimal, and for Stretch Solar+No Lunar it helps the value of Mastery rating get pretty close to the value of Crit rating.

• Hamlet · 09/15/10, 6:00 PM

Fixed the above, will be in next time. I just made Eclipse uptime 1 for one-sided rotations--you wouldn't refresh a DoT during the little gap without Eclipse anyway.

This does make Solar-only look optimal again. Interestingly, this little process clarified the reason why--maximizing Solar uptime as much as possible squeezes the most damage out of our best spell: Insect Swarm.

---

It's hard to know what to make of the information that Solar-only is theoretically optimal. As I keep mentioning, the point of the sheet is more to understand patterns and relationships between spells--the literal output is only exactly relevant when you're working on single-target stationary DPS for more than a minute at a time. Importantly, for example, if there's movement going on, Lunar suddenly becomes important again because it buffs Moonfire. And if there's AoE, you need Solar for Hurricane/WM. Hard to say right now how this will all work out between practice.

iamrelevart · 09/16/10, 12:36 AM

Why would you want to include time spent casting spells that do not benefit from Eclipse in an Eclipse Uptime value. I'm building a model in excel to look at the amount of time in the rotation spent casting spells that benefit from Eclipse and I am specifically avoiding counting any time where I am casting a spell that does not benefit from Eclipse. This includes DoTs.

In my model, I noticed what you're talking about with the glyph of starfire. as far as prime glyphs go, that really does throw a pickle into things. Perhaps a rotation that fully extends each moonfire while swinging Eclipses (meaning you would extend a solar Eclipse by casting Starfire, possibly eeking out two Starsurges in that phase) would work to maximize the glyphs, buffed IS uptime, and still proc Euphoria enough to maintain mana? I'm spitballing here of course, but that seems more in line with the goals that GC seemed to want...

Also, while Solar-only looks optimal because of IS and the way we would be milking it, GC already said that if we settled into that as the best rotation, they would break it. I gotta say I'm with Graylo on the fact that they're gonna do everything they can to make us switch Eclipse procs.

On an unrelated note, has anyone taken a really close look at Mastery? I did so napkin math for my blog today and came to the conclusion that it is our worst stat, itemization wise. Perhaps I'm missing something important, but that really bums me out if that's the case. It's like getting a new b.b. gun and discovering that your slingshot fires off pellets faster, harder, and further.