Friday, March 25, 2016

PSR adding numbers into our subjective raider rankings

At the beginning of HFC I did a little spreadsheet that tracked some metrics for our raiders. Directly or indirectly this objective look at our team helped lead to an improvement in our raiding. At least I believe it did. Once we completed heroic content I dropped the charting, which to be fair was cumbersome to maintain, and we entered mythics almost as a lark.

I honestly didn't expect us to do so well in mythic and I certainly wasn't expecting a Gore clear as early as it came. With things going better but still some performance issues it seemed time to bring out the PSR scores again and see where our raiders stood.
I do want to take a few minutes to explain the metrics chosen and how the PSR score is calculated but first I should note that this is simply a way to collect and compare numbers, it does not take into account more subjective things like group dynamic or other nontangible qualities. It ain't perfect.

PSR stands for Participation, Survivability, and Ranking. I also try to keep track of average DPS but that isn't used in calculating an individuals score.
The scoring is out of 300 possible points, 100 points coming from each category.

The easiest metric to track is to simply see who was there and who was part of each kill. I've weighted the numbers so that being present is 9 times as important and being part of the kill. This is done so that those who show up but have to be benched we can track that and we can track whose not at the full raid. At the same time those who are there on a wipe fest night are penalized too much for missing a load of kills the night before.

Kills are totaled and averaged then multiplied by 10. Similarly nights are added and averaged and multiplied by 90. These two numbers add together to give your participation ranking.

WCL has a nifty little chart that shows the average survivablility for each raider over the course of the night. The way it works is by finding the first time you die and checking what percentage of the fight that was. So if you die 10s into a 100s fight thats a 10% survivability. If you don't die thats 100%. Those who die quickly during a called wipe will have a very slightly worse score than those who take longer to die but this is fairly negligible and averages out over time. This number is averaged from all kills and wipes for the night and then I again average it over all raid nights that the raider was present. Because of this someone who shows up to one raid could have a relatively high S score but of course their P score would be much, much lower so it works out.

Survivability is perhaps one of the most important metrics that defines a raider. Deaths occur for a variety of reasons but ignoring mechanics and standing in stupid  lead to the highest number of deaths. If you can't count on someone being alive throughtout the fight then have an uncertainty in the mix that is not as highly valued as someone who may do more damage....if they live.

The ranking score is a compilation of each DPSers overall ranking for a kill plus the weighted ranking and the ilvl ranking. Each kill is tracked and then the three categories get individually averaged. Finally all three averages are added together and divided by three giving the R score.

The overall ranking looks at all logs from a particular spec regardless of ilvl and compares the raider to that list. Previously with heroic content this could be tricky as it grouped in all raid sizes which can get complex but with mythics set size of 20 the only real change is number of DPS in relation to the healer/tank count. Now it should be noted that this late in an expansion there is the problem of gear creep where enough secondary stats are available that the top players begin to do more experimental builds and push the limits. This is why we also look at ilvl limited rankings so that you are least given a similar playing field with which to compete on.
The addition of weighted rankings is supposed to normalize for those people who meter whore while ignoring the real targets that need downing.

So while this information is handy for raid leads and for forming a 20 man team out of hopefully more players than we need there is also use for it by the raid members themselves. Using S and R scores players can see how they are doing and the more diligent can then begin to search for ways to improve.
Of course it is always acceptable to ask as well.


No comments:

Post a Comment