Basic thing is as sais above: rating are relative. They do not measure you chess skill but how well you win other players that particular pool. And what sort of skill level maps to what sort of rating is then dependent on whole lotta things
- how initial rating is formed. USCF/Elo use initilization based some amount of games while glicko-x used some pre-determined value with high rating deviation estimate (
chess.com 1200 and lichess 1500)
- how strong rating pool is compared to you.
Chess.com/lichess rapid pools are lot weaker than blitz pool. so that alone explains abotu 200-300 points of difference
- and to a minor degree the adaptation algorithm. Glicko type algorithms do not inherently produce different number but there is a small difference due to differerence id convergence speed. They all produce number where 200 point difference means that in long match stronger player win 3/4 of the points and weaker wins the rest.
So in short actual numeric value is meaningless. it only to find you a matching opponent and all the systems work well enough in this respect