It's an interesting point. To test to see what the relative values of scaling were, I poked around Simulationcraft and let it run for a few hours (25,000 iterations - which makes the scaling stats computations take forever). Anyway, the end result was that haste scaled fairly linearly. Mastery and crit eventually became nearly equal. But, ultimately, haste was significantly better than any other secondary stat (once hit/expertise capped) by a wide margin (even in the range of 1000 haste more than now) - for combat anyway.

This suggests to me that there's some weirdness going on with (unnecessary/nonexistent) haste breakpoints in the Shadowcraft engine. As a test, I ran the optimizer and found it suggesting agi instead of haste. Normally, I'd probably accept this. Except the EP spreadsheet I maintain with the values (derived from Shadowcraft) of the weights suggests the previous gem setup (haste primarily in yellow sockets instead of agi/haste. And agi/haste instead of agi in red sockets) is correct. Making it even weirder, after I optimize it in SC, I run the optimization for gems again and it reverts back to the previous setup. That seems suspicious to me.

Yea im also having it not show dps calucations nor stat weights. Which messes all reforging and optimizing

Edit: 03/21/13 - 3:32AM. seems to be working fine now.

I did as Seliathan suggested and changed my gems one by one to agi and agi/haste gems and haste was catching up to mastery slowly untill I got to a gem that changed drastically the EP value for haste as Ondersjaak said to the same values that I posted on my previous message

I actually spent another 3 or so hours yesterday trying to find breakpoints around 7.2k haste and couldn't find a single point in the code near that area where GCD capping was suddenly a concern.

I'm going to need more information, otherwise I'm going to chalk it up to the predicted DR for haste rating. Things like determining how many actions of what per second you have during AR+SB with and without 4piece, etc.

It's intentional, Expertise has a lower EP than other secondary stats as assassination. Mainly because dodged specials refund energy and combo points, dodged envenoms still apply the envenom buff, and poisons are on melee hit cap (so expertise capping is less important than hit capping.)

If you'd like, you can reforge for expertise cap to gain a bit more cycle stability in exchange for a bit lower theoretical max DPS.

Steps to reproduce the bug I am seeing:

Load my profile in (Angosia@Stormrage).
Click optimize gems. (That looks good at first glance.)
Click optimize gems again. (Lower than the first optimize.)

It reverts back to what was setup before (what's on the armory). This is confusing. If it is a downgrade, it shouldn't recommend doing anything.

I know why it's doing this, but I am unable to determine the proper resolution.

Stat weights without gem changes:

Agility 2.711
Yellow Hit 1.872
Expertise Rating 1.667
Haste Rating 1.079
Strength 1.050
Mastery Rating 1.012
Attack Power 1.000
Crit Rating 0.964
Spell Hit 0.425
Hit Rating 0.425
Mainhand Dps 4.334
Offhand Dps 1.380
Mh Expertise Rating 1.395
Oh Expertise Rating 0.272


Stat weights with gem changes:

Agility 2.722
Yellow Hit 1.534
Haste Rating 1.506
Expertise Rating 1.321
Mastery Rating 1.139
Strength 1.050
Attack Power 1.000
Crit Rating 0.985
Spell Hit 0.434
Hit Rating 0.434
Mainhand Dps 4.319
Offhand Dps 1.370
Mh Expertise Rating 1.042
Oh Expertise Rating 0.279

Edit to add: Plugging in Simulationcraft's weight stats says that my gem choices (as indicated in the armory) are better (slightly) than the optimized version suggested by Shadowcraft.
Edit2: Fix grammar and add some clarification.

I've searched through the thread and could only find someone bringing this up with no response so I'll try again: Is the ridiculously high dps gained from the Combat t15 4pc a bug with the engine? Shadowcraft at the moment is finding the 4pc to have an EP of over 25000 for Combat.

Now that's partially because if you have the t14 4pc active your SB uptime is much higher so the EP estimation is enormous. But even if you equip 3 pieces of t15 and no t14, that 4th piece of t15 does calculate as a 20k dps upgrade over a 522 offset piece. The GCD reduction and pseudo-AR during SB is obviously going to be strong for Combat, but that number is quite a bit past reasonable for a set bonus.

Reasonable number? I'm seeing the full upgrade from 4-pc 517 t14 to 4-pc 522 t15 to be around 4k dps total for Assassination and 16k dps for Combat. Pretty much entirely due to that bonus.

Don't confuse poor reforging with fault in the math. That reforger wasn't even optimal in Cataclysm.

It's expected to be that good. AR+SB+4piece is a ridiculous number of CPGs, finishers, and RB procs within a short time span.

Indeed. Consider that in T15H gear, Rogues are supposed to be pushing over 200k. :o) (And we're still not at the top.) The set bonus is the least of our problems and probably the only reason we're propped up in the tier. I'd see this as less of an issue with set bonuses and more of an issue with spec scaling that the tier bonus is patching but ultimately not solving.

The reason that set bonus is awesome for combat is because of the GCD reduction in 4 set. Together with AR glyph we have a 0.5 second GCD during shadow blades + AR and 0.8 second during AR. You just get so much more done with your cooldowns thanks to that set bonus.