Summary so far: Mastery ranges from roughly 25% better than crit to 8% better than crit during shadow blades + killing spree, with the 8% near the high end of player armor and the 25% near the low end. It is around 35% better than crit with JUST killing spree, where armor is not a factor. This is because the extra damage and accurate from shadow blades autoattacks is aided more by crit, with mastery just getting a mild frequency boost under blades..
This season I'm messing with combat PvP. My favorite rogue spec was always HARP at 70 and the objectively niche SHARP at 80, but at 85 combat couldn't get prep, and was trashcan. In MoP, combat could finally get prep, but overall I would say sub overshadowed it entirely.
And sub kinda still does- the numbers show sub rogue way outnumber combat. But unlike in the past, I think that combat actually CAN succeed now (I've tried it several seasons throughout all seasons, but normally had to go sub past 2k). With the likely ability to poach someone from oqueue, I suspect I can push later in the season... maybe.
Anyway, that's my theory, and that's why I'm making this thread: to discuss how to get that big killing spree. Also to discuss strategy if anyone wants to- it's rare enough that combat isn't the kind of well covered space as sub is.
Killing Spree is the preeminent Combat kill button. It's got a small cooldown that gets smaller when you evis or rupture (or I think tempest). It hits hard and fast, and renders you immune to CC. It also keeps you on the target, no matter where they go (which can be good or bad). You can /cancelaura to end killing spree early, but that's about it. You can't spree during anything that makes you lose control of your character: fear, charm, disorient, stun. You can't spree when disarmed. You normally want to land spree with your cooldowns up and in a good insight (yellow or red). If possible, shadow blades should be on during it. All damage you do during killing spree is increased by a small amount (I think like 20%?), which isn't really something you handle with strategy, it's just more damage. Because killing spree uses both weapons, you want two slow weapons. Most rogues seem to pick Prey on the Weak, and if you can long stun a target who can't trinket it, then you should consider that as well.
So mostly you will killing spree when you want a kill. Killing Spree is composed of the following:
Seven main hand Killing Spree attacks.
Seven off hand Killing Spree attacks.
Two or three main hand melee attacks.
Two or three off hand melee attacks.
Normally one to four main gauche attacks.
Normally four to eight wound poison procs.
Remember that offhand attacks are only 87.5% as effective as mainhand attacks, and that autoattacks have a +19% miss rate compared to yellow attacks.
So, my math is saying that Mastery is my best secondary stat during this. If you are running around with reasonably similar PvP gear to mine (you have 522 weapons and some other pieces, but still many 496, having capped every week but maybe not capped much more than 2200), then we can probably round your crit percent to 20%, and your mastery percent to 30%, and your white miss at around 20% (likely a couple less, but rounded).
600 crit rating buys 1% of crit, and that same amount buys 2% of mastery.
Everything in the above table is based on weapon damage, which is a combination of rolling damage for the weapon, then adding in 2.4 times the AP, times a number (I recall dividing by 14, who knows)! The exception is wound poison, which I'm going to ignore. It shouldn't make a very large difference, but feel free to disagree if you think it does in practice.
So the damage will be:
crit times [ (killing spree) + (auto attack) + (main gauche) ]
>We expect to get 13.125W damage out of the spree itself, counted as 1W for each of 7 mainhands and 0.875W for each of 7 offhands.
>During the spree, we will assume you have four autoattacks- two of each hand. You could have three if you are lucky, but my weapon speed under slice is damned near 1.5, so two is a very solid average. You will expect to miss 20% of the time with these. We'll count these as 1W for each mainhand plus 0.875W for each offhand, each times 0.8 for the miss chance. That will give us 3W.
>Assuming your mastery is roughly 30%, those seven spree mainhands can each trigger one, as can the two autoattack mainhands. Each mastery hit is 1.2W. The autoattack mainhands only can trigger when they hit, so we'll multiply their chance to 0.8. This gives us 0.3 * (7+0.8*2) = 2.58 main gauche hits. Each hit is 1.2 times weapon damage, so 3.096W, which we'll round to 3.1W.
So our damage, assuming the 20% crit and the 30% mastery and the 20% miss chance with white hits, is:
1.2 * [ 13.125W + 3W + 3.1W ] = 23.07W
Pretend, for the moment, that we have our choice of 5% crit (3000 rating, roughly within reach of gems, enchants, reforges) or 10% mastery (also 3000 rating).
Crit rising to 1.25:
1.25 * [19.225] = 24.03, an increase of 4.2% total damage.
Mastery rising to 40%:
This changes the mastery number. We up the first number from 0.3 to 0.4: 0.4 * (7+0.8*2) = 3.44 main gauche hits, and multiplying by their 1.2W damage we get about 4.2.
1.2 * [ 13.125W + 3W + 4.2W ] = 24.39W, an increase of 5.7% total damage.
This means that mastery is like 35% better than crit during the spree!
What about haste? Haste does do damage, especially during the rest of the fight. But I don't exactly know how to calculate the average gain during the spree. What I do know is that it will only effect the white swings during the spree (this will in turn give you a bit more expected main gauche, but remember that you have like 9 hits, 7 of them from the spree, upping that to 10 hits is strong but it doesn't change the main gauche like it will the rest of the fight). Unlike crit, both haste and mastery can give you energy back. My intuition is that haste's contribution during spree will be very close to white hit rating- aka, not very much at all. The mastery boosts an effect that triggers way more often than during your normal rotation. The crit boosts everything you do evenly, but this is a big ramp of damage. Haste takes a lot longer to play out.
So, that's math. Post if you disagree with my method here.
What to reforge to for the biggest spree?
Mastery seems bigger than crit, which seems bigger than haste, which is in turn likely similar to white hit.
And lets add consideration outside of spree:
Haste would be top in pve by a lot, but suffers from lack of uptime. It's still good though. Mastery normally trails haste, and crit trails that by a lot.
So my guess is:
For a BIG SPREE:
Mastery >> Crit > Haste ~= White Hit
And you'd probably want to prioritize Haste way over white hit, given how much it helps the rest of the fight.
I'm currently stuck trying to figure out whether it's worth trying to go ALL IN on mastery- for instance, a piece with Expertise and Haste, with the Haste the larger value- I could get more mastery out of the haste, but what if I then have to reforge haste to expertise on another piece as a result? I'm not sure of how much to weigh that haste, but I'll try to guess at something.
EDIT: If you do this during blades (and you normally will), I think that eliminates the miss chance, reducing white hit's benefit to zero, but changing the formula a little, and reducing mastery's edge very slightly. The white hits are, instead of 3W, 3.75W. The mastery changes slightly as well, becoming Mastery * 9 * 1.2W. Some things hit harder as well, but that isn't important for the stats much (unless wound is a bigger factor).
Base:
1.2 * [ 13.125W + 3.75W + 3.24W ] = 24.14W
Crit rising to 1.25:
1.25 * [20.115] = 25.14, still about 4.2%.
Mastery rising to 40%:
1.2 * [ 13.125W + 3.75W + 4.32W ] = 25.43W, an increase of 5.4% total damage.
This makes it around 30% better, not 35%, I think. I don't think it changes the conclusion though.
Updated: If you do this during blades on an opponent with no armor, the above holds. Because shadow blades lets just your autoattacks ignore armor, they become a bigger part of the formula. This shrinks the mastery edge. At around 50% armor (holy paladin, roughly), you will only find mastery 8% better than crit, with a progression up to 25% on higher armored targets.