Heroic is also hugely skewed by incredibly short kill times, low add health, and raids of guild reclears / boost runs with extremely high ilvl.
The reality is no data set is useful without context or explanation to supplement it.
This also isn’t how that works. As soon as a fight has 50 recorded kills public then all parses are added to the data set, it doesn’t matter if within those parses there are only 2 mm hunters. It just means your 100-50th percentiles are all the same log.
Marks is 3rd best? WTF? Why is everyone playing BM then? Why are some people still complaining? Where MM/BM is, I would be (and am) completely satisfied. If anything, im upset Ret is towards the bottom and I feel like its smashing everything.
Ret has weak single target and while it has great AoE, it requires it to lose so much single target to spec into it that it’s not particularly justifiable outside of mythic+, and even then it’s mostly people padding overall.
I feel like this needs explaining a little more. The number of parses has a huge impact on how effective percentile measurement is. For example if we look at a popular DPS spec this season like Assassination rogue, we see thousands upon thousands of parses. The top 5% of those are good players but it’s still a huge number, and there’s likely a big jump in performance between top 5% and the min maxers / people able to pad getting the absolute highest numbers.
Contrast that to a spec that only has a few hundred parses like mm, the top 5% is only a few dozen people, likely the best players in the world on that spec. Meanwhile were that spec as popular as Assa, those same players would make up less than a single percent of the data.