A Saint, on 24 July 2021 - 03:54 AM, said:
Well, i did it. 1 day of trying and here you are, skill rating cange without correlation with match score!
I'm 7-th in machscore. And arrow? As red as beetroot!
P. S. My bad, guys, did two screenshots, but lost one, with arrow. Ill try to get another one, next time.
P. P. S. Did no teamdamage this match.
TL;DR I have looked at over a thousand games. PSR outcomes can be disconnected from our in-game-performance expectations.
The PSR outcome (formula) is based off of the average of your team and the average of the match. You had a PSR change of -2.22. Above 2 is a green arrow, below -1 is a red arrow, and between 2 and -1 is a yellow equals. The formula gives a small PSR advantage to someone on the winning team and a small PSR disadvantage to someone on the losing team.
Your team average match score (the only thing that matters to PSR is match score) is 256.83. The match average match score is 271.54. You were a little above both: 16% above your team average and 10% above the match average. The way the formula works, it slightly weights your performance relative to your team a bit more than your performance relative to the match.
The end result of the formula is that, generally, players on either end of the spectrum tend to see larger PSR changes, and most in the central bulk see smaller changes. Keep in mind that mean (average) is susceptible to outliers. One person can get a really high match score (from damage, AMS, both) and put everyone else at a disadvantage. During events there is a lot of match score and damage farming. This happens with multiple players per match and this affect those averages and, in effect, penalize those that do not farm match score. People sandpapering each other with LRMs, RACs, and AC/LBX-2s, for example, can generate high damage values and not necessarily even kill mechs or win games.
I have attached an image of your game below with the PSR calculations. You can see some of what I am talking about there. These data are ordered by PSR descending, with win -1 being defeat.
I have also attached some data from a game I had a while ago. I use OCR to extract data from screenshots and do calculations (plus other visualizations and analysis. I'm a huge nerd. I get more from that than playing MWO itself and often play just to get more data). In that game, the top match score player had 1439 damage in a COR-7A, and with a grip of AMS. You can see what happened to the PSR changes for the other players as a result. These data are ordered the same way (PSR descending), but with victory 0 being defeat.
Bottom line: PSR is about your match score relative to your team, and your match score relative to the match. It's offset a bit in each direction by victory/defeat. Match score is not always the best predictor of performance (by whatever criteria you assign "performance"). Yes, if you do more damage, kill more, and so on, you tend to get more PSR. You also tend to win more if you do those things. If your team wins, by the nature of most wins, you are all generally scoring well relative to the match average. Things like damage farming and AMS can subvert expected outcomes. You could do really well on your team, but your team gives up tons of damage score to the other team.