Jump to content


Photo

Cataclysm Balance Theorycraft and Spreadsheet Thread


  • This topic is locked This topic is locked
231 replies to this topic

#1 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 01 July 2010 - 01:57 AM

Obviously numbers may change (or spreadsheet might have a number or two wrong), but the spreadsheet seems to show Starsurge as almost pointless for PvE.

Same cast time as talented Wrath, but does less damage than Wrath (@3000 SP), has a lower crit rate, and generates less Lunar Energy. It also doesn't scale as well as Wrath. It has somewhat better mana efficiency, but we aren't supposed to care about that.

If trying to proc Solar, it has lower DPS than SF, and also lower Solar Energy/sec. One thing it can be used for is as a "finisher" to proc Solar Eclipse. If you only need a little energy, Starsurge is a faster cast than SF. You'll often be able to reduce the Eclipse cycle by about 0.7s, and that should be worth something.

It also has a lower crit rate than the other nukes.


Other minor observations:
With 2s talented Wrath, it doesn't seem like NG needs big changes.
MF spam while moving costs 6 mana per cast?

Edit: First post says spell coefficients aren't necessarily correct. Give Starsurge a coefficient of 1.4 (its base damage is about 1.4 times SF base damage), and Starsurge has appropriate DPS for a 15s cooldown.

#2 copialinex

copialinex

    Von Kaiser

  • Members
  • 59 posts

Posted 01 July 2010 - 02:30 AM

I think the time to proc Eclipse is not too short (I loved proccing Eclipse back to back).
These are the times I obtained for obtaining 200 energy either with WR spam and with SF spam only:
Assuming Improved Eclipse after Euphoria & no Nature's Grace.

WR spam:
Posted Image

SF spam:
Posted Image

Notice the huge amount of haste/crit needed to bring "proc time" under 15s (assuming casting DoTs doesn't generate energy & Cataclysm ratings).
The formula I used is: ProcTime=(INT(200/((EnergyPoints+(ExtraEuphoria*%Crit))*1.3))+1)*(BaseCastTime-0.5)/(1+%Haste)

#3 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 03:06 AM

Similar to things I thought while working on it. SS, being unaffected by Starlight Wrath and a few other talents, is almost strictly weaker than Wrath, unless it has huge base damage or an unusual coefficient. Probably best for PvE purposes would be to have it generate more energy--15 is far too little to matter (but then you still have the problem of the cycle being too short).

Moonfire--yeah, well if Moonglow and WoC stack :P . The base DD of Moonfire is very low though. I suspect it will be bumped a lot at some point, or all the Moonfire DD talents will be kind of pointless. WoC making it free is nice, but MF still has to do a worthwhile amount of damage in normal use.

#4 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 06:13 PM

I think the time to proc Eclipse is not too short (I loved proccing Eclipse back to back).
These are the times I obtained for obtaining 200 energy either with WR spam and with SF spam only:
Assuming Improved Eclipse after Euphoria & no Nature's Grace.

Notice the huge amount of haste/crit needed to bring "proc time" under 15s (assuming casting DoTs doesn't generate energy & Cataclysm ratings).
The formula I used is: ProcTime=(INT(200/((EnergyPoints+(ExtraEuphoria*%Crit))*1.3))+1)*(BaseCastTime-0.5)/(1+%Haste)


Main difference I think is that I was using 36% for max rank Improved Eclipse; you were using 30%. We don't really know how it will work at the moment since it's NYI and the tooltips for all 3 ranks are inconsistent (we also don't know if it will affect energy gained from Euphoria). At 30%, it does take a good bit more haste/crit to hit 15s. Also, any DoT refreshing during Eclipse will give a bit of leeway.

I guess the rotation becomes a lot more cumbersome in my mind if the end of one Eclipse crashes into the beginning of the next one. For one, you want a little gap there to refresh DoT's and then cast Starsurge. Proper scaling of haste/crit is also affected if uptime hits 100% and can't increase any further (do we really want more haste/crit caps this expansion?). Also, there's randomness in the proc time due to crits. So if the rotation's very tight, sometimes you'll be lucky with crits and the next Eclipse will start before the current one ends.

Not to mention the problem with lost Eclipse uptime is even worse when you're used to 100% uptime. An interruption during Eclipse will also cause you to have to spend X seconds uneclipsed proccing the next one. It will be a pretty substantial effect proportionally if the cycle is tuned for ordinary DPS at near-100% uptime.


Off the top of my head, a good solution to both make Eclipse somewhat smoother and make Starsurge have more defined role in PvE is to slightly reduce the energy gain from Starfire and Wrath and substantially increase the energy gain from Starsurge. Then Starsurge begins to really feel like a spell you can use to control your Eclipse procs. That's more interesting to play with, and I think one of the design goals of the new Eclipse was to make it feel more controllable (which the current proposal doesn't do very well).

#5 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 01 July 2010 - 06:15 PM

Download the spreadsheet, change the MF strategy to "Always Refresh", and the displayed marginal value of Haste jumps to >30. Move Haste from 300 to 299 (or to 301) and the marginal value is <2.

In cataclysm, there will be lots of places where the marginal value of Haste changes fairly significantly:

1) One more (or one fewer) DoT tick (this is what happened above, SF-glpyhed MF went from nine to ten ticks).
2) If NG remains unchanged, the point at which NG-Wrath cast time (including latency) drops below 1.5s (significantly boosts NG uptime).
3) Point at which GCD reaches 1s.
4) If NG goes away, Haste has a fairly high value at the point it allows another cast during Eclipse (although I don't think the spreadsheet reflects that, and you don't really know where that point is unless you have very high confidence in your latency number).

I see the spikiness as a potential problem. Sometimes you want to see the spikes and know exactly where they are. Other times you'd prefer a "smoother" behavior. Perhaps the spreadsheet could have a "smooth haste" option that does something like evaluating 1/20*(change in dps when haste changes by 20).


Edit: Eclipse proc rate

We'll be spending 7 talent points, and all of our +Mastery gear boosting Eclipse. To me, that means Eclipse needs to account for a big hunk of our DPS. I'd really prefer a "level" eclipse strategy where we see something like a 30% dps boost, almost 100% of the time, to a "spike" strategy, that is a 60% dps boost, 50% of the time.

Also note that a "level" strategy is much friendlier to movement and DoTs. Lost (or wasted) Eclipse uptime becomes less significant. With the level strategy, Haste still has linear scaling. With the burst strategy Haste has quadratic scaling (it reduces Eclipse downtime and increases casts/second).

Level strategy means 15s (or less) to move the energy bar 200.

#6 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 06:33 PM

Yeah, I think I will have to do something like that. I'm hoping they won't design Nature's Grace in a way that bakes in a reachable haste cap again, but there will always be the issue with DoT's. At some level, DoT's can be handled without changing much. We'll all know the breakpoints (I'll probably add a box to the sheet for computing them based on current talents/buff, just as now).

Squeezing in an extra nuke during Eclipse is an interesting point. In the past, I was never convinced that it was worthwhile to discretize nuke cast times for that purpose. With predictable Eclipse starts and (potentially) no Nature's Graces to randomize cast times, it might become more accurate to do it. I'm not looking forward to figuring out the breakpoints with latency though :/ .

#7 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 07:04 PM

Hmm, so you actually want overlapping Eclipses and 100% uptime? I don't know--for one, you not only need to generate the energy for the next Eclipse every 15s, but also refresh DoT's. Energy generation would have to bump up considerably to do that reliably. And the Eclipse mechanic starts to feel a bit degenerate to me at that point. It's just a bonus for continuous casting.

#8 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 01 July 2010 - 07:18 PM

Hmm, so you actually want overlapping Eclipses and 100% uptime? I don't know--for one, you not only need to generate the energy for the next Eclipse every 15s, but also refresh DoT's. Energy generation would have to bump up considerably to do that reliably. And the Eclipse mechanic starts to feel a bit degenerate to me at that point. It's just a bonus for continuous casting.


Well, it's at least nudging you to rotate back and forth between SF and Wrath. And if they overlap significantly it could drastically lower the cost of movement. Tune it to give you, say, four seconds (or heck, even just 2) leeway after you've refreshed your dots and maybe popped Surge. You get a little buffer to either be allowed some movement, or lower the required bandwidth on when to swap main nukes. Maybe allow ticks of DoTs to generate energy, so that it takes a little setting up to reach that comfort zone. It's not as interesting or intricate as, say, basing your bonus on the amount of energy you have, but it's at least fairly stable, and requires a modicum of setting up.

#9 lissanna

lissanna

    Von Kaiser

  • Members
  • 90 posts

Posted 01 July 2010 - 07:19 PM

Moonkin damage tooltips from spell bar (not actual damage from casting)
Using insect swarm, starfall, & focus glyphs...

level 82 moonkin #'s (150 spell power without gear, no talents):
Moonfire: Burns the enemy for 230 arcane damage then an additional 342 arcane damage over 12 sec
Wrath: (2.5 sec cast) causes 732 to 813 nature damage
starfire: (3.5 second cast): causes 1179 arcane damage
Insect swarm: causes 2208 nature damage over 12 sec.

Level 82 moonkin #'s (3319 spell power geared, no talents)
:
moonfire: 820 instant, then 2058 over 12 sec
wrath: 3002 to 3083 nature damage
starfire: 4358 arcane damage
Insect swarm: 7170 nature damage over 12 sec

#10 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 07:22 PM

Outstanding, thanks. Just check the note I put on your blog while you were posting this--you can get the talented spells too as long as you note what talents you have. The only things that should matter are: spell damage mastery, Earth and Moon, Gale Winds (for Typhoon), Master Shapeshifter (if you're in form, which you probably shouldn't be) .


They were 0 talents, with and without gear (with spell damage #’s recorded). That should give you a good start. Tooltips are scaling with gear, so it’s easy to have accurate tooltip #’s after gear for comparison. Actual spell casting testing takes about an hour to do, since I need reasonable sample sizes for things like wrath that have damage ranges, so I’m not sure I can get to that today.


Tooltips are great, actually. When dealing with talented spells it's a bit trickier since we have no way of knowing what talents they take into account. But based on the info from my unnamed Alpha testing helper, the tooltip did increase when points were added to Earth and Moon and to the Balance mastery. As long as those are true, we don't really need target dummy shooting except as a confirmation.

#11 lissanna

lissanna

    Von Kaiser

  • Members
  • 90 posts

Posted 01 July 2010 - 07:41 PM

The following talented testing with 60/0/13 talent spec: Talent Calculator - World of Warcraft
Using insect swarm, focus, & starfall glyphs.

Level 82, out of moonkin form, tooltip damage #'s, 140 spell power no gear:
Moonfire: 300 arcane damage, 390 over 15 sec.
wrath: 839 to 931 damage
starfire: 1350 damage
insect swarm: 2954 damage over 14 sec
starsurge: 1610 to 2131 damage
starfall: each star does 355 to 409 damage
typhoon: bugged & says 400 damage. Actually does 1831 damage

level 82, out of moonkin form, tooltip damage #'s, 3319 spell power with gear:
Moonfire: 955 damage with 2355 over 15 sec
wrath: 3437 to 3530 damage
starfire: 4990 damage
starsurge: 7197 to 7718 damage
starfall: 1029 to 1083 damage per star
typhoon: bugged & still says 400 damage... actually does 2290 damage

Without E&M on the target, out of moonkin form, the starfire damage on target dummy is same as what is on the tooltip, so the tooltip #'s look accurate this build outside of moonkin form.

#12 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 07:45 PM

Processing Lissanna's numbers for coefficients:

Starfire: 1.00 (unchanged from live)
Wrath: 5/7 (changed only to reflect cast time increase)
Insect Swarm: 0.260 per tick (up from 0.2--sure you didn't have a Glyph of IS slotted?)
Moonfire DD: 0.186 (up from 0.150)
Moonfire DoT: 0.135 per tick (up from 0.130)

e: doing the ones from second post now.

#13 lissanna

lissanna

    Von Kaiser

  • Members
  • 90 posts

Posted 01 July 2010 - 07:50 PM

Had to remove glyphs to repeat insect swarm & starfall testing.

Testing with talents, no gear 140 spell power:
Insect swarm: 2338 damage over 14 sec.
Starfall: 326 to 375 damage per star

testing with talents, with 3319 spell power from gear:
Insect swarm: 8960 damage over 14 sec
starfall: 999 to 1048 damage per star

No talents, no gear 140 spell power:
insect swarm: 1752 over 12 seconds

No talents, with 3319 spell power from gear:
Insect swarm = 6708 damage over 12 sec

#14 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 07:52 PM

The following talented testing with 60/0/13 talent spec: Talent Calculator - World of Warcraft


This shows the damage and haste masteries to be 9.42%, which is what they would be if masteries didn't cap out at 51 points. Can you confirm the mastery bonus in-game is still 8.01%?

#15 lissanna

lissanna

    Von Kaiser

  • Members
  • 90 posts

Posted 01 July 2010 - 07:55 PM

This shows the damage and haste masteries to be 9.42%, which is what they would be if masteries didn't cap out at 51 points. Can you confirm the mastery bonus in-game is still 8.01%?

Mastery gives: 8.01% damage, and 8.01% haste.

#16 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 01 July 2010 - 09:08 PM

The nukes seems pretty straightforward: Starfire and Wrath have standard casttime/3.5 coefficients, but Starsurge has a very high coefficient: 1.535 (so Erdluf was right about that being their route to make SS useful).

Having a bit of trouble figuring out exactly what's going on with the DoT's.

IS:
First dataset (post 10) gives a base of 255/tick (+/- 1) and coefficient of 0.2/tick after dividing out 1.3 for the Glyph.
Second dataset (post 12) is incomplete.
Third (post 14), first set gives a base of 255/tick (over 7 ticks now), and a coefficient of 0.26/tick (1.82 over the full 7 ticks), after dividing out talents. So that's weird.
Third post, second set gives a base of 255/tick and coefficient of 0.26.

So maybe the Glyph affects base damage only? (which would be almost certain to make it . . . pretty awful).

Or maybe the tooltip is not correctly accounting for the Glyph's effect on the coefficient.

MF DoT:
Untalented coefficient is 0.540 over 12 seconds, talented coefficient is 0.540 over 15 seconds (after dividing out talents). Once again, something's not affecting the coefficient correctly, either in reality or just in the tooltip (but again, note that for IS, the apparent coefficient did increase to account for the extra tick, so I don't know).

MF DD:
Again, between talented and untalented table, DD seems to have been increased by 16%, but not coefficient.

Sfall:
Same, w.r.t Glyph of Focus.

Attached is my (kind of messy) computation table.

----

Liss:
--Can you check actual IS/MF damage on a dummy and see if it conforms to tooltips at high spellpower and with glyphs/talents?

Attached Files



#17 copialinex

copialinex

    Von Kaiser

  • Members
  • 59 posts

Posted 02 July 2010 - 12:14 AM

Main difference I think is that I was using 36% for max rank Improved Eclipse; you were using 30%...


I misunderstood the Eclipse tooltip; I thought it was Nature/Arcane , so I take back the "back to back" (Cata frenzy).

I think I didn't make myself clear with the dots issue: If the time to achieve 200 energy is under 15s, it doesn't matter,
you can (must) use the extra time to cast DoTs, the problem is when the time to achieve 200 energy is under 15s-2GCD (or 1GCD if single DoT rotation).

PS. I redid my calculations with a very simple implementation of NG and including the 5% haste buff, they reduced the time needed to achieve 200 energy by an amount I didn't thought of, all of this with Improved Eclipse being 30%, if it is 36%... you can imagine (the 30% data was picked from wowtal.com). Also, I treated energy points as damage buff, first, stack additively, then multiplicatively.
PS2. Could somebody please confirm the relation between Euphoria and SS? I noticed that in both wowtal and wowhead, the first point is said to affect SS, but not the second. Thank you very much.

#18 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 02 July 2010 - 01:11 PM

I honestly wouldn't worry much about Glyphs at this point. With the addition of Medium Glyphs we'll probably see massive changes across the board. And those are probably the last things I expect them to fix up. Plus, I'm curious how much they will effect our game play from a DPS perspective. GC has made his vague insinuations that glyphs will be changing because they don't offer the choice of gameplay they originally were designed to instead becoming a "which glyph is best" sort of deal.

For now, it's probably best to ignore glyphs entirely and wait until we are closer to the release.

PS2. Could somebody please confirm the relation between Euphoria and SS? I noticed that in both wowtal and wowhead, the first point is said to affect SS, but not the second. Thank you very much.

That's a fairly common bug with talents in their early stages. It's safe to assume that Starsurge is still (or will be) affected by Euphoria with 2 talent points spent.

#19 lissanna

lissanna

    Von Kaiser

  • Members
  • 90 posts

Posted 02 July 2010 - 10:30 PM

The moonfire tooltip #'s aren't matching what it is actually doing once you talent into it and have on all your gear. Here's what I have from casting moonfire & insect swarm at the target dummy...

Using the 60/0/13 build, 3319 spell power, 7.44% haste from gear plus haste mastery, out of moonkin form:
Moonfire has 1063 direct damage, plus 783 damage per tick for 5 ticks.
Insect swarm has 1279 damage per tick for 7 ticks

Using the 60/0/13 build, 140 spell power, only haste mastery, out of moonkin form:
Moonfire has 304 direct damage, plus 129 damage per tick for 5 ticks.
Insect swarm has 332 damage each tick for 7 ticks

I think haste isn't being handled well in the tooltip scaling or something strange is happening for DOTs, where their actual damage isn't reflected by our tooltips directly. I'm not 100% sure what's broken.

these numbers are without glyphs, I believe.

#20 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 03 July 2010 - 02:13 AM

Thanks lissana.

Actually, those IS numbers are pretty close to what you'd previously reported from the tooltip. Perhaps IS is now up to 26%/tick to make up for the missing hit debuff.

If you had 7.44% haste from gear, and another 8.01% haste from Mastery, IS should have had eight ticks according to the new rules.

I plugged MF numbers into Hamlet's spreadsheet, but threw in another 1.16 multiplier for DD (BotG and IMF, which I assume are additive with each other, but not with E&M or the mastery bonus - assumptions will need to be tested):

I get direct at 204 base DD, with an 18% coefficient (live at 80 is a 441 base with a 15% coef). DoT ticks at a 86 base and also a 18% coef./tick (Live 80 is 200 + 13%)

I also assume you weren't casting under the influence of the new WoC.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users