Hitting CS during the CS debuff for a chance to proc another CS is like paying $5 to play a game where you flip a coin to win your $5 back, except instead of 50:50 it's more like 10:90. Muspel's right, don't do it.

The default Arms Simcraft ruleset actually prioritizes CS over Slam for the last GCD in a CS debuff if you can't MS, OP, or Execute. Adding that rule doesn't seem to do anything (two 50k iteration simcraft runs, one with and one without the CS rule, gave mean DPS (se) of 39363.2 (10.07) and 39362.0 (10.05) respectively, which gives a p-value of 1.000 for the difference of mean DPS).

This is odd, now that I think about it - the situation where the two rulesets behave differently does come up (e.g. MS OP CS MS OP ?), so why the statistically-indistinguishable DPS? Is there a way to have Simcraft print out a debugging line whenever a particular action is taken?

-----

After more fiddling in Simcraft, it appears that the conventional "use BT on HS, only and always" rule might not be correct.

As before, here are two Simcraft results, each 50k runs, one with the default ruleset (the top image in the sets below), another where we immediately consume all BT procs with HS (bottom).

Summary:

Ability usage:

Ruleset #1, for reference:

Ruleset #2 is exactly the same, with a new "actions+=/heroic_strike,use_off_gcd=1,if=buff.battle_trance.up" line:


Why the DPS loss? Look at the ability usage tables. By using BT on HS immediately we seem to be getting extra HS's (unsurprisingly) but for some reason we're cannibalizing Slams to do it. (This means that whatever's going wrong is doing so before execute phase, since during execute we never Slam anyway.)

Why would consuming BT on HS prevent a Slam? If we're below 35 rage and thus would refuse to Slam without BT (according to Simcraft's ruleset, to ensure we have enough rage for an MS or CS next).

Why would we be below 35 rage on a Patchwerk fight? It'd make sense if we were playing poorly and hitting HS when we shouldn't, but Simcraft's ruleset doesn't do that. It does have this line:

actions+=/heroic_strike,use_off_gcd=1,if=buff.inner_rage.up&target.health_pct>20&(rage>=60|(set _bonus.tier13_2pc_melee&rage>=50))

... which would put us at >=30 rage.

Is Simcraft Heroic Strike'ing too aggressively during Inner Rage pre-Execute, making us lose Slams?

It's pointless to even ask what you're doing wrong if you can't provide logs.

Target dummy tests are a horrible way to measure your DPS as Arms. Without the 10% haste buff (and incoming damage), you generally won't get enough rage to sustain a rotation without being forced to use Battle Trance procs on abilities other than HS.

TLDR: Simcraft testing confirms the conventional wisdom that Mastery > Haste.

Haste > Mastery (although barely) in fights with negligible incoming damage and no exterior haste buff like Nozdormu's, but there aren't any this tier.

It's also not a big deal - you're talking about a change of 50-100 dps, tops.

-----

I've been shifting from Fury to Arms over the past few weeks and have been lamenting the lack of good guides. One sticking point has been reforging priorities. I noticed in passing that Simcraft rated Haste > Mastery, both for my own gear and for the BiS set, but most guides I read + most warriors I talked to + most popular warriors I armory'd were reforging Mastery > Haste. None of the guides I read showed their math, though, and the warriors explained it as conventional wisdom and not something they'd checked the numbers on themselves, so the discrepancy has bugged me for a while. This post is me sorting it out.

How does haste help us?

1) More white damage.
2) More rage gen. This helps two ways:
2a) Usually heroic strikes, since warriors in reasonable gear already have a rage surplus while sustaining their GCD abilities. However:
2b) Fewer missed GCDs. The default Simcraft ruleset actually ragestarves itself with HS and occasionally misses a non-Battle Trance slam (see my post above, same page). More generally, haste makes rage starvation through poor HS usage more forgiving.

If haste helps us through rage gen, though, it's devalued when we're getting more rage than usual, either through other sources of haste (e.g. Nozdormu's buff in Madness) or from incoming damage. I figured that explained the discrepancy, and that there should be some incoming dps breakpoint at which haste and mastery cross over.

Mastery and Haste scale factors in 50k-iteration simcraft runs with the default Arms ruleset and the Arms T13H BiS set, with 0/5k/10k/15k incoming dps (top to bottom), ticking each second:



Somewhere between 0k and 5k incoming dps, the rage gen from incoming damage makes Haste worse than Mastery. Which fights this tier have <2.5k incoming dps? I can't think of any.

A slight complication is that the tests used incoming damage that ticks every second. In a real fight, incoming damage is burstier, so incoming damage as a rage source is less reliable and Haste would retain more of its value. Here are the Haste and Mastery scaling factors at 5k incoming dps, ticking at 5k damage / 1sec (top) and 100k damage / 20sec, again with 50k iterations:



A slight shift, but Mastery > Haste still seems more likely than not.

-----

You can also tell how big a difference reforging Haste vs. Mastery makes. Let's say you can reforge 300 points of Haste into Mastery or vice versa, and you're taking 10k incoming dps:

300*2*(2.0094-1.8151) = 120 dps. Not much, but now we know.

Edit: I'm wrong. Disregard this post.

Opportunity strike does not generate rage.

As a side note, haste will affect Deep Wounds damage indirectly, by giving you more auto-attacks (which can crit) and more rage (for more heroic strikes). I'm unsure if Deep Wounds is affected by armor (I think it is), but if so, then getting higher CS uptime from SD procs will further increase the damage.

Deep wounds is a bleed and it's pretty much common knowledge that bleeds aren't mitigated by armor.

If you were to proc Deep Wounds twice in a very small period of time would that not get into munching?

The way deep wounds works is any crit resets the duration and is recalculated to add damage to the debuff. It doesn't matter how many times you crit in any window of time, the end result is the deep wounds damage is always increased.

At least that's how it (and similar abilities) should work, but I guess some of the fire mages on these forums could tell you a whole lot about their "ignite munching".
Since deep wounds and ignite have the same behaviour for ages (in regards of double dipping), I could imagine the same effect applies to deep wounds, too, but it may be less of an issue.

Correct on both counts.

You're right, my mistake.