Investigating Stat Targets & Minimums
March 14, 2017 Leave a comment
One of the things I regularly see in Discord spec channels is “I have X% Crit, Y% Haste, Z% Mastery, what am I doing wrong?”, usually followed by something like “You should drop some Crit to get A% Haste”. Questions & responses like these appear to be carried over from prior expansions, but are not really applicable anymore due to a number of mechanic changes.
The big one is haste breakpoints. Back in Mists you would want to aim for specific haste targets to gain an additional dot/hot tick, but this was changed to what I refer to as the “Universal Pandemic” system in Warlords and had Partial Ticks added in for good measure. Pandemic was a Mists passive effect for Affliction Warlocks that granted up to 150% of the dot length when it was recast, but lowered to 30% when applied to everyone. The “partial tick” thing only occurred when dots expired, returning a portion of a damage tick based on the time between the last tick & the dot expiring vs the current tick time. These two mechanics did away with haste breakpoints.
So where do these figures originate from now?
Simulating characters is easier & more accessible these days, so combined with a large number of variations in talents, legendaries, artifact traits and so forth it means players are running their own stat weight sims to work out what gear to use. As I’ve mentioned before stat weights are showing how adding more of each stat will affect your output, but many people use them as gear rebalancers and end up in loops where priorities switch back & forth.
A lot of the current targets occur because players running their own sims notice those priority changes and make correlations between the fact that Haste is worth more than Crit with hitting a certain Crit percentage. This then gets passed around until nobody is really sure where it came from or why it exists.
This is the point where I went “I don’t really like this so I’m going to do some analysis”. The first iteration was investigating a haste claim for Arcane, but the larger & easier to use analysis was on Frost Death Knights. For this I ran multiple “actors” in a single simulation, where each actor had 1% more of a stat than the last from 0% to 99%. If you look at the charts you’ll see that while there are spikes & dips all over the place the changes are all fairly linear so that a spike at one point is offset by a similar dip before or after it, while mostly being in a +/- 0.5% error margin.
This is kind of an example for where things aren’t quite so linear. The blue line is the raw DPS gain and has a higher but diminishing gain for each percentage from 5% through to around 20-25%, but becomes fairly consistent after that. This sort of graph refutes any claim of a minimum required value as we would expect to see a large spike if one existed.
As per usual, I’ve included both a sim profile template as well as auto-analysis on additional tabs so you can run this for your own spec or setup.