Tuesday 25 January 2011

Hit me!

Something has been troubling me since I came back to WoW, well lots of things have actually, but one thing more than most; hit caps, or the lack of them. My natural instinct when hitting 85 (if you'll excuse the pun) was to ensure that I was hit capped above all else. When you do some reading, and I've done lots, it would seem that hit is no longer the be all and end all of stats. Head over to shadowpriest.com and you'll see that the stat priority is as follows:

Int: 1.0000
SP: 0.7935
Haste: 0.5031
Crit: 0.4040
Mastery: 0.3857
Hit: 0.3741
Spirit: 0.3732

You'll notice that hit languishes just above spirit and the ginger step child in terms of priorities, now normally this would come with some sort of disclaimer along the lines of "hit calculated after reaching cap of xxxx" but no such statement was made. Was this a mere oversight, is hit still as important as it ever was? It seems not! You might have got an inkling by now that I'm not the type of player who takes things completely at face value, I like to know what's going on under the hood as it aids my own sanity, and I feel understanding the mechanics of the game help me to improve. So off I trotted to various forums and Google in my quest to get some answers, not any old answers or opinion, good hard FACT's backed up with some old fashioned honest evidence.


I have to say I was sorely disappointed; I accept that there's always going to be a bit of a grey area with 'the best' way to DPS because Blizzard don't publish their algorithms for calculating damage done, nor should they, and the fact that everyone's gear, lag, play style and a multitude of other factors will be ever so slightly different. As with any scientific investigation it is possible to develop test theories by modelling observations, which is pretty much what the simlecraft community have done to give us the stat priority.

I had a bit of a Google for material, checked the usual suspects, Elitistjerks.com, shadowpriest.com etc. and the official 'shudder' forums which reminded me why I never go there; I swear half of the people posting there still think the world is flat because "it said so in a book, once". Anyways I digress. In summary, I struggled to find the good hard facts I was looking for, I tried to sift out the mere conjecture and the downright idiotic to give me the next best thing to facts, anecdotal evidence based on observations which have log's or models which hold true in most cases. These theories are not necessarily wrong, or bad for that matter, just incomplete, and in lieu of a complete the best we have to work with.

So in an attempt to concisely answer my question of "is getting to the hit cap top priority" I've made a long and rambling post to come to the conclusion of…."possibly not"… sorry. Here is a summary of what I believe are the key points
  • Anything over 17% is (still) completely pointless.
  • The 17% hit cap seems to sacrifice DPS, stacking Int and haste in favour of hit if you're at the 15% level seems to be the way forward.
  • I suspect, hit is suffering from a diminishing return type rule, if you're vastly under the hit cap then its value to you increases – I've yet to read and digest the simlecraft in its entirety and fully get to grips to say this for certain.
  • I'm really not sure why hit is slightly better than spirit on the priority list, spirit gives mana regen in addition to hit, so surely it should be slightly higher (even if this is nominal) rather than lower? – again I need to do more reading as it's far more likely that I've missed something that the numbers being wrong.
  • The general consensus seems to be to always gem for Int and Haste where possible, unless gemming for something else gives you better overall stats by triggering a slot bonus or enabling a meta-gem.

If it is the case that the hit cap isn't vital, is there a sweet spot to which we should aim? 13-14% would seem to be about there, but this is purely speculation. There's also the human factor to it, people like seeing big numbers on the screen, and hate seeing "MISS" blazing up in front of them, whilst not attaining the hit cap may be optimal, I can see a large proportion of hit-cap-zealots carrying on regardless and casting out the heretics who speak otherwise to their mythical round planet... if there is a sweet spot be it the lowest hit you can attain (there will be a natural minimum for your gear set where you simply can't get any lower), or somewhere in between, the only way we can find out for certain is through simulation and modelling.  

A word on statistics  

The statisticians are lurking at this point, dying to tell you that probability is, by definition, a random effect; what happened last time, and what will happen next time have no bearing on what happens now: if I were to win the lottery tomorrow, my chances of winning it next week would be EXACTLY THE SAME, in the same way if you've just missed with a spell, you're next spell has EXACTLY THE SAME chance of missing, it may miss, and you can consider yourself unlucky if it does, but it will happen occasionally. If you don't believe me try tossing a coin 100 times, did you end up with roughly 50 head and 50 tails? Yes? Did they all come in a nice symmetrical order of Head, Tail, Hail, Tail, etc…? no? that's just what probability does for you, given an infinite number of events it will always even itself out, now infinite is impossible, so we have to settle for statistically significant – if you didn't know for a fact that a coin had a 50:50 chance of landing as a head or a tail, you could work it out by tossing it a few thousand times and recording the results (it will actually, in reality, tend to favour tails very slightly, as the heads side tends to be heavier and is more likely to land face down, but that's by-the-by).

Human nature and intuition is a strong factor, even if it is wrong, if you still not convinced, play a little game with me, it won't take 30 seconds, pick me eight random numbers from 0-9 and write them down on a piece of paper sequentially. Then scroll to the end of this post for some predictions which will astound and amaze you…

My other concern is that if you're unlucky miss on a couple BIG spells like SW:D at sub 25% health or similar then, in a single fight, your DPS will be significantly skewed downward, however your average DPS over, say, 100 identical fights would be better – is this a good thing? Is it better to guarantee(or as close to guaranteed as you can get) a minimum dps for every fight being hit capped, or is it better to have a higher average DPS figure overall? 

So what's my point I hear you say? Well it's this, human beings, by nature, try to look for patterns where there are not. Just look up at the moon, see the face on it? Of course you do, but I guarantee it's not actually a face. Don't mistake your single observations over a short period of time for fact, yes how you play the game is crucially important to your ability to maximise DPS, but when dealing with figures try and take the emotion out of it, look at your logs and always try and prove your assumptions wrong, if you can't you're doing the right thing, even your own logs probably aren't statistically significant keep an eye on the simlecraft and other theroycraft posters, or if you don't like maths, or simply don't understand, pick a few bloggers or forum posters who are reasonably active and you trust (hopefully with time, I might be one of those) and pass an eye over their musings once in a while.
 

So there you are, hit caps, probability and human nature in a nut shell. Simples. I don't really know all the answers (yet), but I'll leave you with a final thought: Don't be a slave to the DPS meters; bosses will go down far quicker if DPS spent the time they do trying to squeeze that extra hundred DPS out of their rotations, instead making sure that they avoid that crap on the floor they haven't noticed which is about to kill them.




Mystics predictions (spoiler alert if you haven't read above yet) 


I'll bet you 20 English pennies that you didn't repeat any numbers in the sequence? If you did I'll bet you're either mathematically orientated in your work or schooling (or you've seen this before and you knew I was trying to trick you), or you repeated only one number? If you asked the same of a truly random number generator, the sequence 11111111 would be just as likely as, say, 92583107 to come out, yet human beings can't tend to manage truly random things.

I'll wager another 20p that some sort of pattern emerges in most cases; along the lines of always going higher than the last number then lower, then higher and so on… or never having two number next to each other which occur naturally next to each other i.e. 0123456789? etc. etc.


No comments:

Post a Comment