Chess.com Rating Distribution

Hi all,

I have been playing for a bit and recently got all of my timed chess ratings above 1000 for the first time. Most of the forum discussions lead me to believe that this means I have finally (or am just about to) hit the status of an "average" player.

Now I have no issue with this, it's all relative and I'm just happy to be improving. But when I look at chess.com's rating distributions, it tells me my rapid rating of 1170 is just shy of top 10% on the site. I believe this distribution is calculated based on players active in the previous 90 days so should not include people who picked it up once and dropped it.

So I suppose my question is, is this distribution accurate? I am flattered to be almost top 10% but that seems a bit surprising given that I have not been taking it seriously for very long. Could also be the case that most people who play on the site just enjoy playing and are not actively trying to increase their rating.

Thanks for your thoughts! 

Will

I've noticed the percentile on chess.com tends to be quite generous compared to other sites such as Lichess

I believe the main reason for that is the high number of beginners and new players constantly joining, which keeps the average low and results in more people being around the middle compared to a normal distribution.

This is especially evident in rapid because 10|0 is the default time control.

Correct. That's why I'm not comparing equal ratings, instead I'm comparing players' chess.com rating vs their Lichess rating (someone with a 1000 chess.com rating and a 1500 lichess rating for example).

The results are that their Lichess rating is almost always higher, yet their percentile is lower compared to that of their chess.com account.

(Keep in mind that my results are from anedoctal evidence only, someone should do a sistematic analysis of ratings and percentiles between the two sites to confirm this).

Percentile is based on math. There's a lot more lower rated players on site, with the average Rapid rating of 664 with almost 51 million players in the pool.

@Martin_Stahl Yeah that's what I said in my second comment.

You are just very talented : )

Yes exactly. Chess.com is a little more "casual" and beginner friendly

I'm late, but this is exactly it. Chess.com playerbase = everybody who plays chess. Lichess playerbase = everyone who takes chess more or less seriously.

Today the average rapid rating is 620 but it looks like the median rating is 400. The average is higher because the high scores skew the average up. This is why Chess.com starts the new members at 400. To see the rating go to Play...Leaderboard ... Stats and you will see the curve.

Starting rating depends on the options chosen at account creation and can be 400, 800, 1200, or 1600

Median rating is not 400.
Also the rapid leaderboard apparently contains wrong data.

Just try adding up all values up to 2300. You'll get 40 million.

2111226+3530727+4268618+4957799+4709930+4258890+3921342+3071589+2326982+1802839+1494672+1079222+751717+531381+381439+278198+189873+127721+81091+66067+34974+16840+7679=40000816

But the table below claims 78 million players. That means we should have 38 more million 2300+ rated players. Obvious mistake there.

Now about median, peak on a histogram is not always equals median. Not in this case. For that rapid leaderboard graph median is somewhere at 550.

If your rapid rating 400 and percentile shown as 50%, then 400 would be median. But first we need to understand what's wrong with the data, why the graph doesn't match the declared player count. Maybe displayed percentile value can't be trusted either.

78 mil players ever played, not including banned or inactive.

The values given for number of players, rank, and percentile have been inconsistent for years. If I divide my Rapid rating by 1-%tile, I get around 14 million active players. That's a far cry from 78 million. If I integrate the distribution curve, I get around 35 million players. Which number is even close to the correct value? Who knows?

I STAND CORRECTED.. the graph is completely off. I suppose it is simply a static image and not linked to the data as it is on LiChess.

There might also be some rating inflation due to the way new players are introduced and ranked, which could influence the distribution and lead to players at 1170 being ranked relatively high.

Someone has done a systematic analysis of ratings but not percentiles. Go to chessratingcomparison.com. You can input your rapid rating on chess.com and it will predict your rating on lichess or fide for rapid. There is a regression program that is based on self reported data. My chess.com rapid rating is 433 and my lichess rapid rating is 948. The program predicted 986 (+- 113) for lichess rapid. Pretty accurate.

Recently a user posted that he reached 50% percentile (median) in Rapid. And his rating was 620 or something.