G'day. I was playing with SimCraft to see if I could figure out what stats -- but more importantly what changes to my rotation -- were good for blood DPS and survivability. I ran into something odd in the results and wanted to run it past the experts and get some feedback because it conflicts with the logical assessment of the rotation.
I used the TMI standard T16N10 boss, and also a custom boss tuned to deliver ~ 100k DPS. I also put at the top of the action list a "cancelaura" on vengeance every thirty seconds, so my vengeance timeline looks a little more like a real fight -- rises to cap over 30 seconds, then drops to zero. Sadly, it basically spikes constantly to a third of the max during the second 30 seconds, then smoothly rises. That *should* be at zero before the rest of my actions, though, and only rise after my "turn" is processed by SimC, if I understand things correctly. (I certainly saw what looked to be effects of this crude simulation of on and off tanking on vengeance.) I did include effective AMS soaking in the simulation.
I started testing DPS and TMI-based scaling to make sure my basic methodology was sane -- because we have a pretty solid idea which stats are good for improving DPS, and which for damage spike mitigation, so if I got the same results out of SimC I could be reasonably confident that my rotation, etc, simulation efforts were at least approximately working.
...and I did. I saw that, eg, crit was one of the best DPS contributors for both, and that mastery/stamina were the peak for TMI reduction; everything behaved as expected in the simulations and I got my confirmation that I could generate the same results ... with two oddities.
The first one was that SimC suggested that lifting my expertise from 7.5 percent to 15 percent was a huge win in terms of DPS. My best guess is that this is reflected in the damage-per-execute for soul reaper with high vengeance, since that was hitting for 600k+ -- and I understand it is one of the few things where you might not land it with only 7.5 percent expertise. I guess missing a few of those *huge* DPS contributor DOTs could be a big loss.
Does that sound reasonable? Is this, instead, perhaps a SimCraft bug where they don't correctly implement the "cannot parry/dodge" property of some of our attacks, so it looks like, eg, more DS, RS, or HS hits land even though they really don't?
The second was that dodge came up as a great way to improve my DPS -- higher weighted than parry. I usually sit at ~ 33.73 percent parry and 8.19 percent dodge unbuffed (before the boss parry/dodge reduction). My best guess was that the parry percentage was high enough that the DR slope for it had actually started to bite, and I saw higher value per point of secondary stats put into dodge despite parry-haste due to SoB and runic procs. Credible?
Expertise does certainly increase DPS. Essentially, raising hit or exp by 1% does the same thing as adding 1% of crit; it increases the average damage of your abilities by 1% (additive), with the exception that there's a cap on its benefit that is different for different abilities.
When you take into consideration that 1% of crit costs 600 rating, and 1% of exp/hit only costs 340 rating, you see why hit/exp are the most valuable per-point DPS stats up to the 7.5% caps. When you also consider that melee hits, Soul Reaper, and Heart Strike reasonably account for 20-30% of our damage, and that more melee hits results in more RP generation from SoB, it's reasonable that expertise past the 7.5% cap is valuable for DPS. Past 7.5%, it should be worse per-point than crit, but similar to or possibly better than dodge/parry, depending on Riposte uptime.
For your second question, diminishing returns on dodge/parry don't affect DPS a whole lot, because the underlying ratings that provide the crit rating aren't diminished, it's only your actual percentage chances to dodge/parry that are affected. Riposte uptime isn't going to be appreciably benefited by the small extra dodge/parry chance you get from trying to balance the DR curves, and the extra melee hits from parryhaste will always be beneficial, whereas the extra RP generation from avoiding very slightly more attacks will sometimes go to waste. I'm not sure where the point is at which you gain more RP from the extra avoidance from balancing DR than from the extra melee hits from parryhaste.
Edit: To put the above point into perspective, let's say the DR curve is so severe that adding 1000 dodge gives me 1% dodge, and adding 1000 parry gives me 0.5% parry. I took just about 100 melee attacks in my last General Nazgrim fight. So, for 100 attacks, I get the following benefits from each stat:
1000 dodge = 1% Dodge:
-1 more avoided attack = 10 RP
-750 crit from Riposte
1000 parry = 0.5% Parry:
-0.5 more avoided attacks = 5 RP
-750 crit from Riposte
-0.005 * 0.24 * 100 = .12 of a melee swing, which gives me 1.2 RP and a little bit of extra damage from the swing itself.
So I get the same crit from Riposte either way, I avoid half of an extra attack with Dodge, and I gain 12.7% of a Rune Strike from the Dodge (assuming 100% of the SoB procs will not go to waste) vs. 12% of a melee swing from Parry. In practice the DR curves will never be that severe. Bottom line is just go for Parry over Dodge.