Spiritual Guidance: The calculus of shadow priest hit in Cataclysm

NOTE (Feb. 8, 2011): With patch 4.0.6 now having gone live, I recommend all raiding shadow priests cap at 17% (1,742 points of hit).
Every once in a while in my heroic PUGs, people get talkative. That can be a good thing or it can be a bad thing. It depends entirely on the person talking.
See, I've learned a lot of good information about bosses and strategy from complete strangers. I don't play a warrior, so it never occured to me that disarming Forgemaster Throngus would make the encounter so much easier on the healer. It was a pretty good day when an impatient healer showed me a shortcut around the Lockmaw trash in the Lost City. And it took me far longer than I'd like to admit to notice the slipstream teleporters in the Vortex Pinnacle.
But for all the good advice, there's been just as much bad advice. The worst of it came from a healer who said, in a copy-and-paste that he likely shared with every random group he was thrown in, "Cataclysm heroics are hard. If you are not hit-capped, you do not belong here. Please drop group, re-gem, and re-enchant until you are at the cap and re-queue."
That's bad advice for shadow priests on multiple levels. Managing hit caps in Wrath was easy -- get to 263 (or 289) points of hit and you were done. Cataclysm is a whole different animal, though. There are a number of important hit cap numbers to know. And getting to the ridiculously high new raiding hit cap doesn't improve your DPS. But you should probably cap out anyway. Unless you're running heroics.
Shadow priest hit in Cataclysm is a complex beast. Let's ... discuss.
The basics of hit
Before we jump into the more complex theory, let's start with the facts. We have a natural desire to have our spells to connect every time, all the time, against all opponents. It's not fun to see a spell miss, and some of our buffs are dependent on successful casts.
Healing priests don't need to worry about hit, for the most part -- all helpful spells require exactly 0 points of hit. For shadow priests, though, whether or not a harmful spell connects is dependent entirely on your opponent's level. There are three types of opponents that shadow priests should be concerned with:
- Other level 85 players Capping out your hit against other players in PvP is a rather easy exercise. To hit another character at level 85, you need a mere 4% worth of hit -- that's 410 points -- to guarantee a hit every time as a caster in PvP.
- Cataclysm heroic bosses Bosses in Cataclysm heroics are level 87. You need 6% worth of hit, 615 points, to never miss against the likes of Baron Ashbury or General Umbriss. All considered, it's fairly easy to get to this point. Between hit and spirit, a random set of ilevel 333 gear should get you to 615 almost automatically.
- Skull-level bosses This is the level raiders need to be concerned with. The bar shoots up dramatically: You need 17% worth of hit (a whopping 1,742 points!) to guarantee against missing Argaloth or any of the other Cataclysm raid bosses. You also need 17% worth of hit to guarantee against missing the end boss in a Cataclysm heroic, but that's not something you should lose sleep worrying about. Crafting a heroic set around the 1,742-point bar is foolishness.
Chasing the holy grail
In the days of Wrath of the Lich King, hit was the holy grail -- the stat. Before you did anything else regarding your gear, you made sure you were at the hit cap. It was clearly worth more than all other stats, right up to the hit cap. Obviously, everyone took that information to heart and carried it over to Cataclysm, thinking that to be constant.
But here's the rub: Early theorycrafting for patch 4.0.3 just didn't support that information. Even today, the theorycrafted numbers feel suspicious. Shadowpriest.com, for instance, tells you that if intellect is worth 1, then a point of haste is worth 0.8 and a point of spellpower is worth 0.7. It doesn't put a value on hit -- it merely tells you that hit is the most valuable thing ever and you should cap it. All of this contradicts earlier information that valued spellpower ahead of haste -- and ahead of hit.
Compounding the confusion: All this information comes from the same damn program from the same 4.0.3 build. If I wanted to settle the pseudo-power debate right here and now, I'd wind up having to use the same theorycrafting program, and that wouldn't solve much of anything. So we're going as basic as possible to answer the question: by loading up Recount and heading to the training dummy.
The great hit cap experiment
Admittedly, my effort to measure the importance of being at the hit cap is less than scientific. Raid fights involve movement; training dummies don't. Still, I tried to be as methodical with it as possible. It certainly won't be the final word, because I can't run 100,000 simulations over the course of a minute the way a computer can, but still, I figured it would prove informative.
Here are the ground rules of my simulation: I'd take a shot at the training dummy 20 times in total. The first 10 times, I'd take my standard ilevel 346 heroic gear and reforge as far away from the hit cap as I could -- basically, the way I reforge to run heroics. Every spare point of spirit or hit was reforged to haste (preferably) or crit (if the gear already had haste). I'd then use this gear to go from full mana to out of mana at the dummy, unbuffed, and take a look at my DPS the moment I went out of mana. To limit the amount of RNG influence, I decided to pop Archangel and Shadowfiend only once per run. I tried to time every cast the way I normally would in a boss fight, and I tried to keep my methods as similar as possible for each run.
Here's my first set of data. At 889 points worth of hit (that's 8.32% to miss a skull-level target):
9,579 / 8,995 / 9,128 / 9,121 / 9,544 / 9,178 / 9,502 / 9,227 / 9,120 / 9,917
Dropping the highest and lowest results and averaging the remaining eight data points gives us an average of 9,300 DPS at 889 hit.
Now, with that information collected, it was time to hit cap myself and take another 10 tries at the dummy. I started by undoing all my reforging, which restored the hit and spirit numbers of all my gear back to their original levels. This only brought me up to 1,409 points of hit (still 3.25% to miss), so I went a step further. Reforging all my mastery to spirit (or hit) was a no-brainer, since every simulation seems to show mastery remains our weakest stat. That brought me to 1,637 (1.02% to miss), just shy of the target. A little more reforging allowed me to hit my target exactly: 1,742 points' worth of hit.
So, back to the training dummy to see what difference being hit-capped made. Here are the results at 1,742 points' worth of hit, with 0% to miss a skull-level target:
9,378 / 9,650 / 9,117 / 9,321 / 8,881 / 9,229 / 8,850 / 9,700 / 9,299 / 9,621
Again, dropping the highest and lowest results, we get an average of 9,312 DPS at 1,742 hit. Surprising? Going "all in" on hit provided very little benefit in the numbers.
Curious, I ran one more set of dummy runs with a hit rating somewhere in between the two -- 1,336 (13.04% hit chance):
9,349 / 9,411 / 9,326 / 9,353 / 9,301 / 9,459 / 9,636 / 9,576 / 9,749 / 9,657
... which works out to an average of 9,468 DPS at 1,336 hit.
Interpreting the numbers
In the above data section, we found that there was virtually no difference in DPS between a hit rating of 889 and reforging to reach the hit cap of 1,742. And we also found there was a slight DPS benefit to having a hit rating somewhere in between -- a hit rating of 1,336 proved consistently (though marginally) better than the other extremes.
Why is this the case? Well, possibly because stat balancing is important. Those pseudopower numbers you can get off sites like Elitist Jerks and Shadowpriest.com are interesting, but their use is somewhat limited. If you have a set of equipment without any crit, then adding a point of crit will be incredibly valuable to you, perhaps more so than the alternatives. Alternatively, if you have a set of equipment that's overloaded with crit, your next point of crit won't be nearly as useful as a point of haste or even a point of mastery. In summary: Pseudopower numbers work, but only if you've got a well-balanced set of stats.
Consider what was happening in the final days of Wrath. All us shadow priests were carrying around somewhere between 263 and 289 points of hit. But at the same time, we were carrying around over a thousand points of crit, haste, and even spirit (which used to contribute to our DPS). Our stats were imbalanced to the point where each point of hit was sickeningly valuable.
Putting together a somewhat random set of ilevel 333 and 346 gear will likely give you some semblance of stat balance, much more so than in the final days of Wrath. With the old (way-too-low) hit cap gone, we can carry around a more balanced amount of hit, which naturally means that hit is worth less. That's what my testing shows. The early pseudopower numbers were probably more right than wrong -- hit simply isn't our most valuable stat anymore. But the numbers say more than that -- pseudopower numbers naturally change when you drastically alter what stats you favor and what ones you don't.
A hit rating of 889 represents an artificially low amount of hit for a shadow priest of my average item level. A hit rating of 1,742 is artificially high. There's nothing magical about the hit rating of 1,336 (13% hit), and perhaps that's why it works the best.
Choosing a hit rating that's right for you
If you're reading this article and thinking, "Well, I'm going to go out and get exactly 1,336 points worth of hit just like Fox," then you're missing the point. Raw DPS isn't everything. The difference between 9,300 DPS and 9,400 DPS really just doesn't matter.
I mean, it matters. Wipes do happen when a boss hits 20,000 health, and a few extra points of DPS can make a difference. But those few points usually don't matter. More often than not, they're just statistical noise.
What did matter between those attempts was that attacks missed. That part's pretty obvious, but missing has consequences beyond DPS. Mind Blast has an 8-second cooldown. If MB misses, you've missed a chance to refresh Replenishment, almost guaranteeing the effect will drop off before you can refresh it. That hurts the raid. Ugly.
And beyond that, missing costs mana. If a cast of Devouring Plague misses, you've lost about 5,000 mana over the course of a single global cooldown. Those misses add up with a low hit rating -- you find yourself running out of mana incredibly quickly. Or, at least, much more quick than you would at 17% hit. Or even 13%. Never missing means your mana usage is far more efficient, which is something that doesn't show up in the raw DPS number.
Beyond that, missing with a cast is a jarring experience, at least to me. Raid DPS is supposed to feel smooth, and a good shadow priest that never misses can plan their next moves in advance. When you start missing, you have to revisit certain spells when they miss and devote extra time to monitoring what DoTs are on what enemies. A shadow priest at the hit cap knows that once they press that Devouring Plauge button, the attack will connect and you won't have to cast it again for a solid 20 seconds. A shadow priest that's under the hit cap has to pay attention to whether that attack hits or not, because if it misses, you need to keep pressing that damn button until it connects.
For shadow priests that have no interest in raiding, the math is simple. If all you care about is PvP, then get to 410 exactly. If all you're doing is running heroics, your minimum hit should be 615. You'll miss on the final bosses, but that's okay -- your other stats (like haste) will be competitive enough with hit that it shouldn't matter too much. And besides, that final boss represents maybe 2% of the time you'll spend in heroics.
For raiding, though, we find ourselves in an interesting situation: Reaching the hit cap can hurt our DPS (slightly), at least in this stage of the game. But ultimately, it's better off for your team, at least if you're a raider. And you want to be a team player, don't you?
Filed under: Priest, (Priest) Spiritual Guidance






Reader Comments (Page 1 of 5)
Rimshot Dec 29th 2010 4:14PM
As a healer, with mana problems in raids:
Listen to this guy. Don't miss your Mind Blasts. Replenishment is nice for us.
Thank you.
Lynxplus Dec 29th 2010 4:17PM
Mayhaps your numbers for hit and non hit builds were the same because you were using a target dunmmy as your target? My guess is the numbers would have been drastically different on an elite, or skull mob/boss/whatever/.
I'm shadow priest who recently upped my hit from around 8% to 14% and have noticed my damage in groups pretty much doubling. I frequently came in last on damage even in non heroic cata dungeons. As soon as I upped my hit% my DPS is significantly better, I'm now first or second on meters...
Randy Dec 29th 2010 4:28PM
Why would training dummies be different? Explain that one since there is a skull level raid based training dummy for him to test this on. Granted at this point there is no movement or need for defense....but seriously it shouldn't be that different.
Fox Van Allen Dec 29th 2010 5:00PM
I did make sure to use the raiders' training dummy each time. It's skull level.
theRaptor Dec 29th 2010 6:56PM
@ LYnxplus
If upping your hit doubled your damage it was because YOU WERE PLAYING BADLY. If you are used to Wrath where Spriests hit capped in blues (due to 6% hit from talents) then you probably weren't paying enough attention and noticing when spells didn't land (I had the same problem in wrath when I tried to run a few points under cap rather then massively over due to gearing issues).
For all DoT classes* hit is less valuable then DD classes because missing 1.5 seconds of 20 seconds of damage isn't a huge issue.
However for raiding in general CONSISTENCY is more valuable then ABSOLUTE performance which is why you go for the hit cap.
* Feral druids for example completely ignore it because they get 80% energy return for misses.
Ez Dec 29th 2010 7:21PM
DPS is higher with good hit because you are actually hitting the mob. If a spell misses there is no dps from it. I believe that is what the moral to the story is.
theRaptor Dec 29th 2010 10:40PM
@Ez
Except the most you can (over an infinite number of casts) miss is 17%. Therefore hit can never account for more than a 17% increases in DPS unless you are playing wrong.
The reason hit used to be so valued was because point for point hit (up to the cap) gave more of a DPS increase than other stats. That is no longer true for many classes due to the new ratings and mechanics.
Animaneth Dec 30th 2010 6:34AM
@Randy: Raid Boss Level training dummies, do behave different than bosses (besides the obvious not responding to attacks or crushing your squishy self). My main is a mage, I usually test specs to decide wich will work better in a raid, frost can't be tested. Frost priority "rotation" requieres you to use Deep Freeze each time it's off cooldown and you have Fingers of Frost up. The reason is that DF does damage instead of stunning to targets that are permanently immune to stuns (bosses, and some elites).
But the training dummies even the raid bosses one are stunnable, so you dont get to see the real benefit of DF.
t0xic Dec 29th 2010 4:20PM
I'd be interested to know if there's any difference between the target dummies and a live raid boss. I think I remember at least one time in the past when the dummies didn't behave like actual raid bosses.
I could be mistaken (I don't have a forum link or anything).
Pyromelter Dec 29th 2010 4:29PM
The problem with dummies is that they aren't fully debuffed, with things like Ebon Plaguebringer, critical mass, etc, and as well you aren't fully buffed with things like +10% spellpower, et al.
Extra crit, for example, is extremely potent at lower levels of crit. Going from 5 to 10% crit is a massive buff, whereas going from 25 to 30%, it's still a big buff, but not quite as massive.
That's why people use spreadsheets and simulators to test dps as opposed to hitting dummies.
Aldarion Dec 29th 2010 4:53PM
They didn't behave like bosses in terms of armor points a long while back. But that doesn't have to do anything with hit and/or spell damage.
Hit percentage is the same and was the same.
Pyromelter Dec 29th 2010 4:21PM
Here's an argument for the hit cap: What if there is a boss with a phase change, where you need to refresh your dots, and won't be hitting the boss for a bit? I'm not currently raiding, but to bring back a couple of ICC bosses, think of Putricide in the phase transitions and Sindragosa before she flies in the air, or the lich king when he heads back to the center of the room.
A random miss in those encounters could really wreak havoc with your dps. And as you noted, missing a mind blast, especially in a 10-man raid without a ret pally, could be devastating to your healers. Another factor in missing is that that is another piece of information that you have to track, either with your dot timer addon or with a combat text notification.
I know a lot of stats get gimped when you reach for hit capping, but I'd take a 1% dps loss at hit cap and feel secure in knowing all my spells are landing, my healers (and myself) were getting replenishment, and that I won't have to worry about tracking missed spells.
Angus Dec 29th 2010 7:24PM
Looking at this, he was showing how in 5 mans it isn't the god stat.
Raids, it is probably the most important stat, but in 5 mans, meh.
Saeadame Dec 29th 2010 7:34PM
Resto druids also give replenishment now, and they get it from their cheapest HoT (Lifebloom) so it will, pretty much, ALWAYS be up since it costs only about 1% mana or less. Plus, Frost Mages, Destruction Warlocks, in addition to Ret Pallies, also give replenishment. I doubt that you will often be in a raid (especially, say, a 25 man where you usually have a few people of each class, and almost definitely a resto druid) that doesn't have at least one of those in addition to you, the Spriest. Missing with a MB isn't as crazy crucial as you think.
Although the DoT thing is a very good point. And having hit cap in raids is probably also a very good thing. But missing on refreshing replenishment isn't as big a deal as some people think it is, there's more specs with replenishment than you think.
Eisengel Dec 31st 2010 12:42AM
True enough, although honestly if you're collecting 333+ gear, you'd almost have to try not to get 6%. Spirit and hit is on a lot of gear. I've noticed the Justice point gear often has Spirit on the healer version, and hit on the DPS version, so no matter what you're getting hit.
Necromann Dec 29th 2010 4:21PM
Thanks for the article Fox
PS, I names my hunter's pet after you, although he's lvl 13.
Necromann Dec 29th 2010 4:24PM
Named* why do the s and d keys have to be so close together.
Priestess Dec 29th 2010 11:05PM
So I can steer.
zhaharik Dec 29th 2010 4:27PM
If you're going to conduct your measurements based on going all-out until you go OOM, then you should be measuring total damage done, not DPS. Your raid DPS is measured from the start of the fight until the boss goes down, not until you run out of mana.
If hitting with key attacks keeps your mana up, then I'm guessing your low-hit parses lasted less time than your high-hit parses?
Fox Van Allen Dec 29th 2010 5:25PM
I alluded to that in the final section, but I guess I can spell it out right here:
Yes, you will run out of mana far sooner with 8% hit than with 17% hit.
I've long been an advocate of "total damage" numbers meaning more than damage per second numbers, but far too many people focus on and prefer to read DPS numbers. Had I just posted raw damage numbers, people would be asking, "yeah, but what about DPS?"
I can't win either way. Still, the final message of the column is the same, and should matter more than those silly little numbers in the sections before.