So, in workweek bustle I skimmed what Paul originally wrote and assumed the "formula" would be something like Trevelyas'
5-year-old suggestion to score players in context with each other's performance each match, then apply that over time to the entire population.
Now I'm wondering if Paul will literally remove PSR increases for anyone on a losing team, and call it "zero sum."
I mean, yes,
you can correlate player skill to W/L rate given enough players and enough time.
But there's a better metric to measure player skill, and that's: player skill. I don't know where this notion that high matchscores can be rigged or "scummed" by certain playstyles indefinitely, because
data shows that it's impossible. Players level off at an average of 250 per match, way too low for consistently high-scoring games with low-skill-floor playstyles. Any loophole players are statistically insignificant. In other words, players who try to fake it end up playing with players they can no longer fake it against. But see, if they happen to be on the winning side and play every single day...
Unless low-matchscore winners are also penalized, it seems wishful thinking that PSR and tiers will end up much differently from the XP bar.
I don't know if it's unchecked belief in Dave Sirlin or whatever, but W/L is not a magical end-all-be-all stat. If it were, the Reds wouldn't have paid $64 million to Nick Castellanos, because they'd just check the Tigers' W/L and assume he can't be good if his team loses all the time.