Saturday, November 17, 2012

Hit is Actually Your Worst Stat

At least when you're deciding which gear to use.  The gist of this article is that when you're looking at gear, most sources will claim that gear with Hit (or Expertise) on it is far better than it actually is.

Note: the following is written with casters in mind, though the same logic applies to expertise.

If you're casual about the game, you've probably seen something that looks similar to this:

Intellect > Hit > Haste >  Crit > Mastery

The primary stat and order of the secondary stats may differ, of course.  If you're a bit more worried about the numbers, you may have seen something like

Intellect: 1.00
Hit: 0.70
Haste: 0.50
Crit: 0.40
Mastery: 30

 or even

Intellect: 3.4
Hit: 2.38
Haste: 1.7
Crit: 1.36
Mastery: 1.02

All three of these formats are guides to what stats your spec desires to perform optimally.  This sort of thing appears on the WoW class forums, Elitist Jerks, sites like AskMrRobot/Icy Veins/Noxxic, and more.

And they're all lies.

Well, that's a bit of hyperbole.  It would be more accurate to say that if you use these stat values, you'll probably pick inferior gear and perform worse.  Let's do some math:

To understand why this is so, imagine you have a choice between two items that are identical besides the type of secondary stats (and we'll use the values from the second guide, the one where intellect equals 1):

Item 1 has 100 hit and 100 mastery

Item 2 has 100 haste and 100 crit

The question I pose to you is, which item should you prefer?


Okay, have your answer?

You probably picked item 1 (and if you didn't, kudos to you).  The easy math is to do the following:

100 * 0.7 (hit value) + 100 * 0.3 (mastery value) = 100 intellect for item 1

100 * 0.5 (haste value) + 100 * 0.4 (crit value) = 90 intellect for item 2

Item 1 is clearly the equivalent of 10 intellect better, right?  Well...not quite.  To understand the problem, let's look at a full gear set made up items with the stats of item 1 and item 2 respectively.

Set 1 has 35% hit and 35% mastery.

Set 2 has 35% haste and 35% crit.

Hopefully you already see an issue: we have way too much hit in set 1 that's completely useless.  So let's do some reforging for both sets.

Set 1 has 21% hit, 14% haste, and 35% mastery

Set 2 has 35% haste, 21% crit, and 14% hit

Let's do our math again for the overall value of these stats:

15 * 0.7 (hit value) + 14 * 0.5 (haste value) + 35 * 0.3 (mastery value) = 28 total value for set 1

35 * 0.5 (haste value) + 21 * 0.4 (crit value) * 14 * 0.7 (hit value) = 35.7 total value for set 2

Whoa.  Set 2 is 27.5% better than set 1 in terms of secondary stats.  But why is that?  We followed what Simcraft and all of these sites said.

The answer, in a nutshell, is that any hit over the cap has a value of 0.  That's obvious, you say.  What perhaps isn't as obvious is that as a result hit is only as valuable as the stats you have to reforge away to get the hit.  It's a matter of opportunity cost, a concept in economics.

Because there is a limit to how much hit you need, the value of hit is that of the stats you trade for the hit.  Another example to illustrate the idea.  You have three items and need to chose a trinket (which gives you four items total).

Items 1 and 2 have 15% haste and 15% mastery.  Item 3 has 7.5% haste and 7.5% mastery.

For item 4, should you take a trinket with 15% haste or 15% hit?  You aren't hit capped (in fact you don't have any hit), so 15% hit seems like the obvious choice.

Okay, let's run the math...

37.5 * 0.5 (haste value) + 37.5 * 0.3 (mastery value) + 15 * 0.7 (hit value) = 40.5 total

But what if we reforge away the mastery to hit on the items and take the haste trinket?

52.5% * 0.5 (haste value) + 22.5 * 0.3 (mastery value) + 15 * 0.7 (hit value) = 43.5 total

Huh.  Even with not wasting any hit in either case, the haste trinket option is nearly 10% better.  If you look at the math, you notice that if we take the hit trinket, we effectively lose 15% haste to keep 15% mastery.  Which is clearly a bad trade since haste is better than mastery in our example.

In short, the value of hit (normally 0.7) in that last example is actually 0.3, since we're giving up mastery for it.

Hopefully you're convinced by now that this logic is right (don't be the guy in the image).  But what does it actually mean in practical terms?

Well, at the beginning of an expansion, it's unlikely you'll have enough of your worst stat (out of haste/crit/mastery) to reforge away to hit.  And even if you could, you'd avoid items with that worst stat on them anyway when possible (if mastery is your worst stat, clearly a haste/crit item is superior to a haste/mastery item).  So in general, when trying to figure out the overall power of an item, use the *worst* of those secondary stats as the value of hit.

With the stat values we've been using, that means we'd value hit at 0.3 (the value of mastery).  Which, lo and behold, would lead us to choosing that haste trinket over the hit trinket.  In short, unless you severely lack hit, you want to be reforging to hit as much as possible and avoiding items with hit as possible.  In particular, trinkets with hit (such as the Flashfrozen Resin Globule) are actually terrible unless you cannot reach the hit cap without it.

But perhaps you're not comfortable trying to do all these numbers and adjustments yourself.  I have good news!  Some places with gear lists (such as AskMrRobot) let you enter in your own custom stat weights.  So instead of accepting the default

Intellect: 1.00
Hit: 0.70
Haste: 0.5
Crit: 0.4
Mastery: 0.3

swap it to

Intellect: 1.00
Hit: 0.40
Haste: 0.5
Crit: 0.4
Mastery: 0.3

The only difference is you adjust the value of hit.


If you do this and check item item upgrades (AskMrRobot actually works very well for this), you'll have a much more accurate upgrade list when trying to figure out which items are optimal to use.

Happy DPSing!


  1. Good point. Adjusting the stat weighting for Hit/Exp will result in a better overall result, but only if the tool was blindly following the weightings, not if it was already taking into account the fact that any value over the Cap is useless. If it was adjusting for the Cap then you're doubling your own work.

    I disagree on Hit and Expertise are the worst stats. They are great until you reach the caps, then they are almost useless.

    The nerdy formula should be written as:
    Hit to Cap > Exp to Cap > other stats.

    Mr Robot does this well enough, as does the in-game mod Reforge which automates the calculation of useful reforging.

    I say almost as sometimes a melee will be standing in front of the boss rather than behind, which means that a little more Expertise can help avoid a parry/dodge/etc. But that is essentially not worth worrying about in your gear config. The fights are tuned expecting the caps to be present, and not exceeded.

    Also you are better off being a tiny amount under the Hit/Exp cap than being over it at all. This is because a 0.2% under will only affect one or two swings/casts in a raid encounter, and the rest of the swings/casts you are doing in the raid will be increased by the alteration of the other preferred stats up by 0.2.

  2. I wasn't as clear as I should have been at the end of the post (I added a sentence at the beginning and clarified the ending).

    In terms of raw DPS value prior to Hit/Expertise caps, Hit/Expertise are typically your best stats, yes. And when reforging you definitely want to reach those caps.

    But when you're choosing gear itself, you generally want to avoid them.

    If your raw DPS value for stats is Hit = Expertise > Haste > Crit > Mastery, then using those values might mean you'd take a Hit/Mastery item over a Haste/Crit item. But because you want to reforge Mastery to Hit/Expertise (and then reforge Crit to Hit/Expertise if you're still under the caps), you want to take the Haste/Crit item instead. Because of the same reasoning that you want to use the Haste trinket over the Hit trinket.

    This behavior occurs because you only need so much Hit/Expertise.