On the other hand haste has the downside that its benefit is being granularly delayed. What I mean by this is:

If you got 5% haste from your gear do you do ca. 5% more autoattack dmg and get 5% more rage no matter what? (forget about glancings, flurry and enrage uptime etc ... this is not the point here).
5% haste meand for every 20 swings you make you get 1 bonus swing. Now comes the crux: If you are not continuously hitting the mob for at least the amount of time it takes to make 20 swings (lets say 45 secs) you are in danger of losing the whole benefit of these 5% haste. Had to jump out of a void zone every 45secs? you just lost part of your haste unless you did it exactly agter a swing connected.

The joust nature of PvP e.g. (being kited, slowed, rooted, CCed, ...) makes all the haste which apllies to autoswings only (i.e. haste for the warrior) most of the time completely useless. Yes I know that this is dependant on all the haste you have (including flurry effects), and it is too simple to reduce it to 45 secs chunks. The point remains that the granularity of the haste benefit (being 1 added swing) is an implicit disadvantage of haste compared to something like crit or hit. I'm not talking about probablity here either. Probability effects like +hit or +crit even out over time. Haste however makes you lose part of its rating every time you have a few seconds of "swing down time" for whatever reasons imaginable.

in short:
1% hit on average IS 1% more white damage
1% haste is always less

Yes Im overlooking the added flurry benefits Landsoul has spoken of, and the benefit of haste to enrage uptime but on the other hand haste "eats flurry" and scales with glancings. Hit is not affected by glancings, benefits flurry and enrage.

I think your movement theory may be flawed. We can't assume how long you have to stand there for haste to be of a benefit. Haste could be the very thing that lets you get an extra attack in before having to move. (example you needed exactly 30 seconds to get your 19th and 20th white attacks, but since you have no haste and had to move @ 29.5 seconds you missed two swings of damage.) Al'akir heroic comes to mind.

Pvp is not the topic nor the logic being used.

By popular request:
Reforging of haste to hit at the 372 BiS level

What you see is that
a) hit is always better than haste
b) the more hit we have, the smaller the advantage of hit over haste (until the point where we reach the hit cap)

Disclaimer: Things might be different in non-patchwerk fights and I won't guess about classes that I haven't played intensively since the last major patch.

Berthold, do those graphs take into account cooldown stacking? If it's using the default action profile(which is mine), it doesn't. I've always been unhappy with the results of cooldown stacking within simcraft, it seems like I should be able to get more than ~100 dps out of delaying FoA to line up with DW. Either way, seems like that would affect your mastery graph.

As for haste vs hit, suicuique does bring up a good point that more haste gets you more glances, more hit gets you more hits.

And while I do agree with the math behind Landsoul's comparison(barring leaving out glancing), I can't help but feel that it's a bit myopic. Things like suicuique's granularity, or more hit means your rage income is steadier. How do you fit those into equations? Lastly, how do you reconcile the difference between Landsoul's stat weights vs Simcraft's?

Landsoul, couple questions about your equation:

1.2833 is the haste synergy, 0.81 is the connect rate, what is that 1.09375? That looks like the adjustment for haste rating getting a worse conversion than hit rating, so that you can compare them directly. Shouldn't that be 0.9375 instead?

I'm assuming the second line is hit. Where's the 0.91 coming from?







P.S.: spreadsheet?

Any chance of getting those graphs for SMF too? That'd be much appreciated.

With our latest release of SimC everyone can do exactly what Bert is doing with results customized to their gear set.

Apologies, but the word "upcoming" in his post and my inability to find this option in 406-14 led me to believe this wasn't available to the public yet, hence my asking.

I'm particularly interested in Haste and Mastery for SMF. They're SMF's two worst stats (with SimCraft consistently showing them near each other at ~0.3 of Strength), yet there seems to be a possibility of reforging the slightly worse one (Mastery) to the slightly better one (Haste). Currently this has always resulted in negligible gains for me. I was hoping the graphs would shed some light on this.

Well, after thinking about it again today I realized my mistake. the 1.2833 multiplier should appear in both numerator and denominator of both ratios. I tend to over-think things and make them more complicated than they are. Sorry for being a little myopic, but it is extremely hard to avoid because full consideration requires glances and crits as well as other little things. The quick equation I was going for was a scalar for Melee DPS when adding 1200 rating of haste or 1200 rating of hit. I will leave out values which are common in both equations this time. Interestingly enough, when you favor glances and crits+DWounds it likes haste more than hit at a certain point.

0.24 = glance rate
.75 = glance scale
0.20 = crit rate
2.48 = crit scale w/ DWounds
(1-(.27-.08))-(.20+.24) = .81-.44 = 0.37 = hit rate at special cap
(1-(.27-.18))-(.20+.24) = .91-.44 = 0.47 = hit rate w/1200 rating
.09375 = haste from 1200 rating

With 1200 haste rating, you simply are getting 9.375% more attacks, so the scalar ratio of NewDPS/OldDPS is 1.09375, and this scalar is also the increase in rage gen scalar.

Now instead with 1200 hit rating the ratio
(.47+.20*2.48+.24*.75) / (.37+.20*2.48+.24*.75) = 1.095602294 gives the relative scaling increase of melee dps
and .91/.81 = 1.123456789 gives the rage gen increase scalar

when adding 600 hit and 600 haste, the scalar becomes [1.046875*(.42+.20*2.48+.24*.75)] / (.37+.20*2.48+.24*.75) = 1.096916826

The 1200 rating comparison favors hit in the case of considering rage gen (2.7% better) and when only slightly when considering overall Melee DPS (0.17% better). Adding 600 hit/haste is better than adding 1200 hit when considering melee dps by 0.12%.



The spreadsheet gives me a value of haste that is 3% better than hit for what reason I haven't fingered yet. The above formulas are only showing a certain part of the picture, though. It's too close to say either way for me. It's almost too close to even matter at all. I promise you won't see a significant statistical difference (with good play) in pure DPS trading hit for haste or vice-versa. Stat geeks will know what I'm talking about.


It works, but sadly not done. Getting everything right is the utmost priority.

Thanks for answering my questions and doing an in-depth analysis. Sorry if 'myopic' sounded insulting, didn't mean it that way. Weird coincidence on that 93.75% coefficient. Anyway, I guess that just leaves the huge discrepancy between Simcraft weightings and yours. I don't have access to your spreadsheet, so I obviously can't make a comparison. And even if I did, not sure I'd really feel like crawling through two different kinds of code.

Just wondering if you had any ideas off the top of your head. Haste and hit are pretty similar stats, can it just be the difference between calculation and simulation?

Simulationcraft does negative delta for hit, so if you look at Simulationcraft Results, you'll see that with -300 hit (the delta used for scaling calcs), the warrior hit rating is dropping down to 500 rating, so under yellow hit cap. You'll have to run it yourself to try a positive delta on hit, which would be a better comparison.

You can easily make SimC run positive deltas by checking the box under buffs. Even with positive deltas hit rates better by a fairly large margin in SimC.

This is the scaling results from my last run of SimC using the BIS TG Fury preset.

If STR=1 then
Hit=.62
Haste=.42