Hacker News Reader

Comment by oldstrangers | Hacker News Reader

1700 on lichess is about 1400-1500 on chess.com. Lichess tends to be about 200 points overrated on average.

Wait, how is that even possible? Are they not using the same ELO algorithm?

They have different player populations. A pool of Elo ratings only rates players relative to the other players in the population.

If you take everyone whose FIDE rating is between 2400-2600, and you create a new pool of Elo ratings for them, their new Elo ratings will range from 1400-1600. If you take everyone whose FIDE rating is between 1400-1600, and you create a new pool of ratings for them, their new Elo ratings will range from 1400-1600. And so on.

Also both sites use the Glicko rating system rather than a straight ELO

https://en.wikipedia.org/wiki/Glicko_rating_system

So what does that say about the relative strengths of the players between chess.com and lichess.com? That chess.com has the more elite players from lichess.com and so the lichess.com players are weaker on average than the chess.com players?

Maybe, but it doesn't really matter in practice. If you play against opponents with a similar rating to you then you'll be evenly matched.

This is a common misconception and the ratings are not comparable across player pools. lichess starts you off at 1500 and so the center of the bell curve is 1500, chess.com starts you off at 1200, so that's where it balances the 50% at.

https://lichess.org/stat/rating/distribution/blitz

https://www.reddit.com/r/chess/wiki/faq#wiki_my_chess.com_ra...

You can't say that based on the ratings. The parameters of the Elo system can be freely chosen, so you can't compare two implementations at all.

Only differences between player ratings matter for Elo Numbers.

You can add 1,000,000 to the ratings of all players on lichess, and it wouldn't change a thing

No, they use different rating algorithms. Lichess is using Glicko-2. Chess com is using Glicko.

Reply on news.ycombinator.com