On the original topic, this is something I've occasionally considered since tier 7 but for some reason I never really did a detailed analysis. Up until last summer it would have been fairly situational due to threat but now threat is virtually a non-issue. Let me therefore attempt to gauge the value of this action.
Let h be our haste, including raid buff. Let c be our crit chance in bear form (typically about 5% less than cat form, due to master shapeshifter and leather specialization).
The cat form swing timer (in seconds) is 1/(1+h), while the bear form swing timer is 2.5/(1+h). We spend 3 seconds in bear form: 1.5 seconds each for the bear form GCD and the mangle GCD (note that we never need to delay mangle since 1/(1+h) < 1.5). Suppose we shift to bear form with time t (in seconds) left on the cat form swing timer; note 0 < t < 1/(1+h). Then 3-t is the time available for bear form swings while 2.5n/(1+h) is the time required to perform n+1 such swings. This means that if 2.5(n-1)/(1+h) < 3-t < 2.5n/(1+h) then we will get n bear form swings. In particular, note that n=1 is only possible for h < 1/6 (<6.1% unbuffed haste), while n=3 is only possible for h > 2/3 (>51.5% unbuffed haste without bloodlust or >16.6% unbuffed haste with bloodlust). Therefore I will assume n=2: we get 2 bear form swings.
I assume 3/3 furor, so we start with 10 rage. Each bear swing generates 16 rage and another 5 on a crit (from primal fury). If mangle crits it will also generate 5 rage. I will assume no rage is generated from damage taken. We now have 42+5r rage available in total or 27+5r rage available for maul, where r is the number of crits in 3 attacks (2 swings and a mangle). If r > 0 then maul is available, so we have a 1-(1-c)^3 chance to have maul available.
Let us now consider clearcasts. If I recall correctly the clearcast chance is normalized against weapon speed at 3.5 procs per minute, so each cat form swing has a 3.5/60 chance and each bear form swing has a 3.5/24 chance to trigger clearcasting.
If a clearcast triggers on the cat swing immediate prior to shifting, I will assume we fail to react to it in time and still shift to bear. However, it is not worth spending this clearcast on a bear mangle so we should abort and shift back to cat. The probability of this happening is 3.5/60 ~ 5.8%, and when it happens the net effect is that we lose 2.5 cat swings and gain 1 bear swing. Furthermore, that bear swing cannot trigger clearcasting (since it's already up).
If a clearcast triggers on the first bear swing, I will similarly assume that we fail to react to it in time and still mangle. This clearcast is therefore lost, so effectively the first bear swing cannot trigger a clearcast.
If a clearcast triggers on the second bear swing, there is the possibility that we might waste it with maul. If it's impossible to react to this clearcast then it's actually probably not worth using maul at all, so let us consider how much time we have to react. Assume maul is used immediately prior to shifting back to cat: there's no reason to use it earlier since it's off the GCD. The second swing occurs at time t+2.5/(1+h) into bear form with t as defined earlier, which means we have between 3-2.5/(1+h) and 3-3.5/(1+h) seconds to react. At a comparatively large value h = 0.3 (~18.2% unbuffed haste) we have between 1.03 and 0.24 seconds to react. This is not a lot of time, and at lower haste values it only becomes worse. One may make some assumptions about reaction time and thereby estimate the probability of wasting a clearcast with this maul, but instead I will simply assume that we do not use maul.
Let
s = cat form swing damage,
S = bear form swing damage,
m = cat form clearcast damage value (typically 1 shred, but one may also wish to include the value of the combo points),
M = bear form mangle damage.
The progression is now as follows:
We have a 3.5/60 chance to abort, in which case we lose 2.5 cat swings (each of which would have had a 3.5/60 chance to trigger clearcasting) and gain 1 bear swing (which cannot trigger clearcasting), so the net gain is S-2.5s-2.5*3.5/60*m (which is negative).
Otherwise we proceed. We lose 5 cat swings (each with 3.5/60 chance to trigger clearcasting) and gain 2 bear swings, where only the second bear swing can trigger clearcasting (3.5/24 chance). We also gain a bear mangle. The net gain is therefore 2S-5s+M+(3.5/24-5*3.5/60)m = 2S-5s+M-3.5/24*m.
Overall, the net gain is 3.5/60(S-2.5s-3.5/24*m)+(1-3.5/60)(2S-5s+M-3.5/24*m). Simplified, this is roughly 1.94S-4.85s+0.942M-0.146m.
Let me try some sample values of
s = 6.9k (4.8k with 24% glance and 47% crit)
S = 8.9k (6.4k with 24% glance and 42% crit)
m = 34.5k (23k with 47% crit, neglecting combo points)
M = 21.7k (15k with 42% crit)
These are pretty close to my own unbuffed values (before armor). Then the net gain is roughly -0.8k (a loss).
Some possible improvements:
- I have assumed we do not use enrage. Obviously, with enrage all bear form damage is increased by 15%. Additionally the extra rage means maul can be used before the second swing, which allows it to be used without risk of wasting the second swing's clearcast. But enrage is only available once per minute.
- If it is actually possible to use maul without always wasting the second swing's clearcast, then it is probably a slight gain to do so.
- It should sometimes be possible to abort the mangle if the first bear form swing triggers a clearcast, depending on timing.
- In the case where the mangle consumes a clearcast, depending on crits it is likely (specifically there is a 1-(1-c)^2 chance) that maul will be available before the second swing, in which case it can be used without risk of wasting a clearcast.
Also note that on Ultraxion, m is suppressed (since shred is unusable) and M is boosted (since mangle is glyphed), which may substantially raise the net damage gain.
Some further effects that I have neglected:
- Fury swipes. This is actually more powerful in bear form than cat form due to its internal cooldown, so it would be a slight damage gain.
- Vial of shadows. The proc itself is only marginally weaker in bear form than cat form, but the slower attack rate means bear form might also delay a proc. This would be a marginal damage loss.
- If we are taking damage then we may gain some vengeance.
Edit: Refined the estimate slightly. Aborting also causes a small loss of clearcasts.
Never heard of this one. How does it work, would you create Savage Leathers in combat?