Rating distribution graph

You learn more when you lose, so lower rated players are indeed better.

If you graphed the curve here youd see a big difference to real life chess like if you graphed the OTB ratings table published by USCF. Integrity is the difference. In real life players must live with their ratings. They also must manage their clock in a way that will not get their face slapped. Online chess is a different world. The USCF 50 percentile falls around 1050. You can google it easily.

That 1050 average includes both kids and adults that are currently active and there are a lot more lower-rated active kids. If the kids and adults are put into separate pools the active kids' 50th percentile is somewhere in the 800s and the (significantly fewer) active adults' 50th percentile is more like 1300-1500 (can't remember where off hand). The 35-45 year old adult average is that few hundred points higher and the rating committee uses the bonus point threshold to keep that average steady. It considers 35-45 year old players to be mature and stable (no longer improving quickly and not yet starting to fade with age) so it uses that as the midpoint. Note that the general increase in chess knowledge does mean that a current 1400 adult knows more and is a better player than a 1980s-era 1400 adult, but ratings are designed as a measure of relative strength compared to other players, not a measure of absolute strength over all times.

The bonus point threshold is where a good tournament result that increases your rating has a portion of that gain doubled so that improving players can have their ratings catch up more quickly to their strengths without taking points from other players. As an example, if the bonus point threshold is 12 then an initial 16 point tournament gain would give 16+(16-12)=20 points and a 50 point gain would give 50+(50-12)=88 points. The threshold has ranged from 10 to 35 over the years while keeping that 35-45 year-olds' average steady even with an increasing number of kids that start out low-rated and rapidly improve.

FIDE does not have the large percentage of rated kids that USCF does, so FIDE has not yet dealt with a need for bonus points. Though if FIDE really does want to start rating kids (with many having three-digit ratings) then it will probably need the overhaul that caused it to back off on rating lots of players in the lower rating levels.

How much elo do you get per game?

The USCF non scholastic 50 percentile is about 1265

So yes there is defnitely a difference between counting every body and deleting scholastic. The meat of my point was. Whether children or not. The people playing USCF events OTB are sincere players trying to play with integrity. Online players try to find inventive ways of messing with Mister Elo's formula without commiting to chess. Thats a big diffrence.

Interesting that I can make the top 5% rapid on this site but can't even get into the top 25% for classical in Lichess.

Just because this can be a pay site doesnt mean it cant be free too. It has way more people at the bottom that give up around 5 or 600 and forget they had account. Thats why i canpossibly be top 6% here.

I'm not sure where you can double-check that. I am in the 99.9 percentile for daily but I have no percentile listed for Blitz or Rapid where I am higher rated but inactive for more than a year. Forgotten accounts might not get included.

6/30/2024 edit dropped from 1941 to 1933 and at #3205 it is 99.8%, indicating about 1.3 million to 2.1 million daily ratings counted (99.75% to 99.85%).

A better explanation for the difference is simply different populations.

Not everyone who has a OTB rating has a chess.com account and vice versa.

While I can imagine ways you can game the Elo system online that you can't do OTB (e.g. challenging people who you think are overrated), I doubt most people can be bothered doing that.

It seems like far too much effort for imaginary internet points, and an overinflated Elo just means your random match-ups will be too hard.

Global rank. Divided by percentile of Rapid. Its counting something like 24 million accounts. IDK if thats everybody or not. But it means 22 or 23 million under 1321 Rapid.

Nowdays(in 2024) the average elo already dropped to 620 elo. How is this possible? chess.com/leaderboard/live/rapid

More members joined that are lower in strength than those that are higher in strength.

Water

I'm 1800 rapid and I sometimes get 65-70 % accuracy just because of some missed wins or opening traps

Martin.

Thanks for posting this. A few days ago, someone posted a similar graph of USCF ratings, so it is easy to compare them.

in my view, the two organizations include very different populations. The USCF (and likely other national federations) has members who are serious players. Whether they play at clubs or in tournaments, they have devoted much time to studying chess. Even the 1600-1700 players at my club have spent hundreds of hours studying the game.

Chess.com includes many—perhaps mostly—casual players who have heard about the game and decide to give give online chess a try. On average, they are much less skillful than the average USCF player. As Metuka pointed out, they are similar to the USCF scholastic players.

👍