Keeping the median ratings at 1500

@Sybotes said in #10:
> If the average level is raising, I will lose rating points in long term even if my personal skills don't change?

theoretically, yes. but that will happen anyway, rating regulator or not. if everyone but you improves, your rating will drop.

> Are you really talking about the median or about the average rating? For statistics, that's quite a difference.

we are talking about the median. the 50th percentile. the 2nd quartile. the 0.5 quantile.

and if you want to be precise: "average" has mostly colloquial usage. some (for example wikipedia) use it as an umbrella term, which would make the median a type of average. you probably wanted to say arithmetic mean.

@glbert said in #11: you probably wanted to say arithmetic mean.
I did. Thanks.

@bluezorro251 said in #9:
> Interesting. The rating is a relative measure. If everyone gets better, no one gets a better rating. There is no objective standard to measure against.

yes and never has been as it is impossible. You are only measured by you win/draw/loss against opponents and that is floating target. But there is no other way. Or actually is finding accuracy of moves correlates strongly with strength. Not the way is computed in lichess though. But that is quite computation intensive so not really worth it. Moves that played when game is no longer contested should not be included. That would work at least above FIDE 2000 quite well. like this paper does the opposite predict error based on FIDE rating chance.amstat.org/2020/09/chess/