That IS clever, and I’ll add it to my list of stuff which influences it
Ok, here is how it works overall.
Lets establish a baseline
You have a bell shaped curve, so, if you are low ranked, then, the players you can have in your games, are weighted towards the higher SR than you side.
If you are above the mid line, then the reverse is there.
If you plot out HOW much the weighting is for each SR, and check against the win rates, you end up with this REALLY nice matching curve.
Which is pretty awesome, because shows that SR more or less works.
Ok… now we have a base, which is your EXPECTED win rate for your SR. (which ALSO scales your win / loss SR from games, you NEED to hit that to break even)
We both know the mirroring effect, which pushes heroes towards THAT mid. for one to win, the other has to lose, but, the winner is usually the higher SR player.
Again, no surprises there.
What IS surprising is the effects of one tricks. They end up placed correctly soon enough, and then start winning/losing games based on the Sr average.
If they were winning more than the average implied, then they would move SR until they get to their status point.
So they PUSH the SR rating ALSO towards the mean for that rank, and do NOT influence the win rates for a hero.
Since they one trick, the attack / defence weighting doesn’t really do much either, since they will be playing on both. (but, I will need to think about that slightly more, and add that effect into the overwatch matchmaker simulation project, to do testing)
So who does? Well, the people who don’t one trick them.
That is where the influence comes from, people who either don’t normally play the hero, or people who main other heroes, but play some of this one.
Now that doesn’t mean they don’t indicate power to some degree. But what it REALLY excels at, is showing how strong a hero is, in the hands of a person who doesn’t know them well.
Or, where the skill floor is.
Which is why seeing someone who first takes up a low win rate hero which they are SURE is op is so funny to watch.
Because by definition they are NOT newbie friendly. The one thing you can say without a DOUBT about the low win rate heroes is they are NOT low skill (which is really low skill FLOOR)
But yeah… Attack / Defence DOES modify winingness. That is brilliant!
It would ALSO push the effects of the “by sr” averages changes, now that I think of it. Time to put it in the simulator!!!