Tuesday, April 6, 2010

New Rage Mechanics in Cataclysm, some numbers

Hi, guys.

So, just for my own edification I set up an MS eXcel simulator to see what the numbers would look like in cataclysm for rage generation as dps.

Most of what I found wasn't unexpected, but I'll talk about it anyway.

To begin with I set up a trial combat similar to what might happen vs a target dummy. I did not take level or damage or anything else into account, since according to blue's those won't figure in anyway. Instead I held a few things that I could modify to run the numbers.

First, I'm assuming expertise cap. Perhaps this is a mistake, but if not, it just makes the misses happen more frequently and I'll talk about those in a minute. Second, I didn't exactly use appropriate battle tables since all I'm looking for is a basic understanding of what will be happening. Third, I allowed user input hit percentage (and I assume you're at the 27% dh miss chance), crit chance, and haste rating to allow for any combination that matters to you. I also allow the user input of blizzard assigned elements such as rage per second from hits (this is the normalized value that the weapon speeds would modify so they all had a standard rps output) off hand modifier (which blizzard states will be 1/2) and crit modifier (which blizzard states will be 2).

Now, to get a solid and clear understanding of what happened, let me give you some of those numbers. I randomly picked 1 as my base normalized rps number. Regardless of what actual value is here, the results will simply be multiplied by the resulting number so that I can talk about absolutes such as twice as much rage as X situation or whatever.

Next I wanted to get some baseline numbers so I could compare haste's benefit to rage vs crit vs hit. The easiest way to do this is to start with white hit cap. Yes, I know that's WAY up there, but that's where I'm starting since I think that's probably going to end up how we have to do things.

Then I started with 0% crit chance or haste (which is not that reasonable) and modified haste up to 100% and then did the same with guaranteed crits every swing. Here is the table of those results for baseline numbers:


So, what do these numbers mean? The first thing I notice is that crit %, in the new system, will translate directly to a multiplier for rage generation, just as it does for dps now. Clearly with numbers between 0 and 100 % crit, there will be variation, but again, these are just baseline numbers.

Notice that with no chance to crit and no haste to speed up your weapon swings, your rps is pretty close to the 1.5 you might expect from the description of how rage is generated. 1 rps from your main hand, and .5 rps from your off hand.

As you add haste, you notice a slight improvement, approximately on the order of a multiplier to rage gained. 10% haste is roughly equivalent to 10% more rage gained, as you might expect. And, clearly, since crit also multiplies, we're not surprised when double rage from crits and double rage from haste stacks to give us approximately 4 times the rps.

I want to reiterate that these are baseline numbers only and cannot be used definitively until Blizzard makes these plans more concrete. So, now I have a fairly good idea of what to start from and I have some clear notions immediately.

1 - As far as rage generation is concerned, haste and crit will work multiplicatively to increase your rps.
2 - Since according to the blue posts so far, haste will do nothing else than what it already does for us, the dps impact of these stats will remain approximately equal.
3 - As far as rage is concerned haste and crit are equivalent in terms of adding, but having both haste and crit values will be equivalent to the product of the two, hence better than the sum.

I also do not yet know what position in this table Blizzard will pick to normalize around, but notice that if your standard rotation (feels good, plenty of rage, but not too much) is normalized at about 3 times the set base rps, people in low gear levels might feel considerably rage starved, and in high level gear (which will now stack crit / haste / hit instead of current stacks of armor pen/strength/crit) you'll be seeing about twice as much rage generation as they've normalized around, and you may still find yourself rage bloated.

Knowing that rage generation is about the same, but also knowing the cost of each point of crit/haste, we can maximize a functional over this set of parameters and see that the optimum ratio of crit / haste (admittedly using current values for these, which may change in Cataclysm)
will be 2/3. So, if you have about 1.5 times as much haste percentage as you have crit, you will optimize your rps.

Ok, now for the secondary side of this.

Note that they still haven't changed the way haste works for dps. Unless they do something about this later, all we'll have going for us in this department is the hope that our white swings will do more damage. As it stands, a good raiding warrior should be seeing white damage in the neighborhood of about 10-12% of his overall dps. With haste affecting nothing other than white swings, this means we really see only 10-12% of the effect from haste that we see in critical strike as far as dps go.

If we factor that in, from a dps perspective, we see our cost for haste skyrocket to somewhere between 273.25 for one percentage point dps increase up to 327.9. At that cost, the optimal ratio for our crit/haste goes to 7/1.

So, essentially, dps would dictate we should budget for 7 times as much crit as haste, but rage generation dictates we should keep a 2/3 ratio of crit chance to haste. This will likely create the dilemma for all of us at the release of Cataclysm, having to decide between a ratio closer to 2/3 for purposes of rage, or 7/1 for purposes of dps.


Now, in all of this, I haven't yet addressed hit rating. From everything I did in tests, any miss chance, be it from misses or dodges (or parries if you're standing in the wrong place) start from those baseline numbers and begin to reduce your overall rps. How much is not entirely clear without some meta-analysis of these simulations, but from initial inspection it seems that the drop is basically equivalent to the miss percentage on average.

So, if you are expertise capped and have 17% hit rating, you can expect to reduce the numbers in the above table by approximately 10% all the time from misses. So, what you see then is a formula something like:

total rps = BaseRPS x (1 - miss/dodge/parry chances) x (1+crit chance) x (1+haste percent)

And I think the lesson I took from this is that every 1% hit you don't have will correspond to a more than 1% rage loss.

So, I expect (and this is completely aside from any values these things may take, and is based solely on their explanation of how rage will work) that we will see a score priority of something like:

hit until white capped, expertise until soft capped > haste, crit

(and the ratio of haste to crit will be up to the individual, but I suspect that it will probably end up at 7 crit = 1 haste, since I figure that raiding individuals SHOULD be able to easily hit the normalization level that Blizzard sets, so they'll focus on the dps instead of extra rage we can't bleed off.)

2 comments:

  1. My god that's a lot of math...but I get it. (sadly)
    IT makes to much sense for haste to help us out on things other than white swings, truthfully I think they could work in some benefit from haste on more than our whites swings, logic tells us (as does physics) that if something is moving faster it's hitting harder.

    ReplyDelete
  2. I agree with you. If haste actually benefited our dps or playstyle in some way so as to make it more desirable, then I'd be feeling very positive about the change. Even if they didn't make hasted swings hit harder, if they reduced the gcd (even if it was down to some minimum) it would make it 'feel' like we were hitting more and our dps would be improving as well.

    ReplyDelete