Wrongs with Rating Accuracy


Game review rating estimation says you played like an 1150, not a 2450
I think it’s important to note that rating estimates are also dependent on the elo of the players involved. Someone who plays a 90 accuracy game as a 1500 will have a much higher elo estimate than someone who plays a 90 accuracy game as a 1000 or 500

I was just checking it and noticed that, too. But it doesn't matter about rating based on who plays because it changes over time, which is weird. Why? What makes it adjust?

I don't pay much attention to rating accuracies or Elo estimates at all.
Unfortunately, they're often incorrect ...

often my average daily game accuracy is around 86-91% and yet it allways tells me i played like a 1800-1900. there might be something of a problem im not sure

...
I think it’s important to note that rating estimates are also dependent on the elo of the players involved. Someone who plays a 90 accuracy game as a 1500 will have a much higher elo estimate than someone who plays a 90 accuracy game as a 1000 or 500
Exactly what I found out when I run the same game through the engine, but tweaking players rating in the .pgn file, the engine would give same accuracy but very different rating estimate for both players:
"1400" player losing to "2100" player, ELO estimate 1050-1900, accuracy 60.6-87.2 respectively
"1400" player losing the same game to "800" player, ELO estimate 900-1600 respectively, same accuracy