Why 600 on Chess.com Feels off? A Data-Driven Look at the Rating Landscape

Sort:
Avatar of lmh50

What an interesting analysis! I should do something similar. In my case I need to disregard games that I resign with less than 10 moves because I suffer from a tendency to resign if my opponent chooses an opening that I don't feel like playing. Disregarding games with less than 10 moves, and looking at just the last ten games (not enough for any significance)

Won 2, drew 3 lost 5

my average accuracy when I win: 70.5 (opponent 59.5)

my average accuracy when I lose: 79.5 (opponent 78.1)

my game rating when I win: 925 (opponent 500)

my game rating when I lose: 1120 (opponent 1090)

my rating over this period: 483 (opponent 493)

The conclusion is that I play much the same and it makes no difference to the outcome (in fact I tend to play worse when I'm likely to win, better when I'm losing - go figure!). Certainly down in my sub-500 area, the accuracy scores aren't too bad. Of my last 10 games, I have accuracies above 75% in six, my opponents in three (six over 70%). This isn't random stuff. The game ratings are a load of rubbish; they don't really reflect the game at all, because they are based on your actual rating, not just the game. The exact same game will be given a different game-ratings if it were played by two different pairs of players.

Avatar of nklristic
Rodrigo-Moraes wrote:

4. Precision Scores
  • My precision: 69.9%

  • Opponent precision: 71.1%

26% of my games had precision below 60%. For my opponents, it was just 17%.

Games above 75% precision made up 38% of the sample for me and 40% for my opponents. That’s high for a rating group often labeled as "beginners who blunder every move."


  • My average game performance: 897

  • Opponent average: 887

Yet we’re both rated around 623. Something’s off.


My point is not that I play perfect chess. My point is that the games — mine and my opponents' — are being evaluated by the platform itself as high 800s or even 900+.

 

So while people critique individual errors, they often ignore the bigger picture. The data shows that many games in the 600 pool are not being played like traditional 600-level chess.


💡 Conclusion

Chess.com’s 600 today is not what it was five years ago.

  • Precision is often above 70%

  • Games are cleaner and more disciplined

  • Over half of the games are played at a level above 1000

 

Thoughts?

I just want to emphasize 2 points here - accuracy and that rating estimation at the end of the games.

The problem with those 2 is that they are taken way too seriously. You are mentioning that 600 today and 5 years ago is not the same. Well, hopefully not, chess is evolving, people should be better a bit. But it is not in the way you think.

Accuracy today and 5 years ago on chess. com is not the same. It is not calculated in the same way. 5 years ago, I had a game with less than 20 accuracy. Today it would be measured around 50-60 (I even shown such an example several years ago). They did it so people would feel better about their games. Today almost nobody scores very low for that reason, so accuracy has to be taken with the grain of salt, especially on lower level where it is more probable that someone will play with very low accuracy.

For instance, a few years ago, when new accuracy measurements kicked in, I had an unrated game with a lower rated opponent. In around 10 moves he blundered 3 times and resigned and had an accuracy around 70. That is completely crazy. One can't take chess.com accuracy as a serious tool for this discussion, especially if they compare it with accuracy 5 years ago when it was calculated differently.

As for rating estimation, that too is just a toy. You said it yourself, it is based on the ratings people have, plus almost always people are overrated with that accuracy. Just how you on average have + few hundred rating, I have many games over 2 000 estimation and I am sure that on average I am several hundred points above compared with my rating.

The only games where I score below my rating is when I mess up early and lose quickly, or when I play a bad unrated game against lower rated opponent and still win in most cases. These type of games are relatively rare, most of the time it is 1 800 + for me, with many games around 2 100 and more.

This metric is made for us to feel good as well. It is certainly inflated. Be sure that actual 2 000 are playing much better than those estimated 2 000 rating games I have.

Avatar of bullet_308

I should have paid attention in math class

Avatar of hiagoskuba

It's trully (1) A philosophical dilemma though:
Real knowledge, skill, wisdom
VS
Need for a general numerical representation and consensus

2) Or even an epistemological issue, that is, how to learn how to learn.

For which additionally one could add: "When, at what point do all these techniques and resourcefulness become a proper art, art of life or living art?"...

Avatar of lmh50
nklristic wrote:
 

I just want to emphasize 2 points here - accuracy and that rating estimation at the end of the games.

The problem with those 2 is that they are taken way too seriously. You are mentioning that 600 today and 5 years ago is not the same. Well, hopefully not, chess is evolving, people should be better a bit. But it is not in the way you think.

Accuracy today and 5 years ago on chess. com is not the same. It is not calculated in the same way. 5 years ago, I had a game with less than 20 accuracy. Today it would be measured around 50-60 (I even shown such an example several years ago). They did it so people would feel better about their games. Today almost nobody scores very low for that reason, so accuracy has to be taken with the grain of salt, especially on lower level where it is more probable that someone will play with very low accuracy.

For instance, a few years ago, when new accuracy measurements kicked in, I had an unrated game with a lower rated opponent. In around 10 moves he blundered 3 times and resigned and had an accuracy around 70. That is completely crazy. One can't take chess.com accuracy as a serious tool for this discussion, especially if they compare it with accuracy 5 years ago when it was calculated differently.

As for rating estimation, that too is just a toy. You said it yourself, it is based on the ratings people have, plus almost always people are overrated with that accuracy. Just how you on average have + few hundred rating, I have many games over 2 000 estimation and I am sure that on average I am several hundred points above compared with my rating.

The only games where I score below my rating is when I mess up early and lose quickly, or when I play a bad unrated game against lower rated opponent and still win in most cases. These type of games are relatively rare, most of the time it is 1 800 + for me, with many games around 2 100 and more.

This metric is made for us to feel good as well. It is certainly inflated. Be sure that actual 2 000 are playing much better than those estimated 2 000 rating games I have.

That's all very true.

The percentage accuracy is an interesting one. If you calculate it as the percentage of moves that are accurate, then quite reasonably a 10-move game with three blunders is 70%, but the point is that one blunder can lose the game completely. There is no point in being 95% accurate if all your games end by hanging a queen. In reality moves aren't scoring fully-accurate or fully-inaccurate, and I notice that the algorithm really doesn't like people who push pawns a lot. I frequently lose against them even though they're only scoring 45% accuracy. On the other hand, if you play a long standard opening or a long logical end-game where there aren't really any possible inaccurate moves, the accuracy score can be vast!

The game ratings are absolute clap-trap. I'm a 500 player and get game ratings up to about 1300, which is the most unbelievable nonsense because I get flattened by anyone above 800, guaranteed. But what's wrong with the occasional feel-good factor? I suppose the best thing is to use them as an internal measurement: was that one of my better games, or worse games?

Avatar of I_throw_in_the_Tal
Petrosian94 wrote:

A certain skill level is required for a particular rating range. From your perspective a group might be underrated, due to a comparison with the past. For now it might be the new normal.

The estimated rating of a game is nothing more than a sales tool (just as the puzzle rating). The platform wsnts to sell something and keep the buyer buying it.

I have had students going from 600 to 1150 in less than 6 months just by putting in some work. If you want to improve the key is to focus on required skill sets. You need tactics and calculation.

Puzzle ratings and estimations don't reflect an "actual elo" or anything, but my estimations are usually within about 400 elo of my rating. And I suck. My games end when either myself or my opponent blunders big enough. I think that's pretty good for some nerdy get rich quick scheme that I have never paid for.

Also, kids absorb knowledge passively. Sometimes I forget how old I am. I couldn't reach 1000 elo in my lifetime even if my best friends were Magnus and Hikaru.

Avatar of Just_an_average_player136
Rodrigo-Moraes wrote:

Hey everyone,

A while ago I posted here when I was struggling to reach 500. Then I posted again when I finally hit 600 but couldn’t maintain it.
Now, I’m hovering consistently between 600 and 660. So I decided to analyze a sample of 47 games I played recently, all rated, and share the numbers behind what I believe is some ELO discrepancies and a shift in what the 600 ELO really means today.


1. No, It's Not Cheaters

Out of 47 games, with 47 different opponents rated in the 600–650 range, only 1 was later banned for cheating. This is not a cheater problem.


2. Winrate by Color
  • White: 50% win rate

  • Black: 39% win rate

I mostly play Petroff Defense as Black (often transposing into Four Knights), and e4 / Scotch Game as White.
So petroff is not working for me hahaha, I think I will start to get sicilian defense against e4.


3. Overall Record
  • Wins: 21

  • Draws: 4

  • Losses: 22

Slightly negative overall.


4. Precision Scores
  • My precision: 69.9%

  • Opponent precision: 71.1%

26% of my games had precision below 60%. For my opponents, it was just 17%.

Games above 75% precision made up 38% of the sample for me and 40% for my opponents. That’s high for a rating group often labeled as "beginners who blunder every move."


5. Move Quality Breakdown
Metric Me Opponent Excellent 1.0 1.0 Best Moves 8.0 8.7 Great Moves 5.0 5.0 Good Moves 4.0 3.5 Errors 1.3 1.8 Blunders 0.8 0.7

This is not the chaos people think of when imagining "600 ELO chess." The games are relatively calculated.


6. Chess.com Game Ratings

Each game is assigned an internal rating score based on how well it was played.

  • My average game performance: 897

  • Opponent average: 887

Yet we’re both rated around 623. Something’s off.


7. Rating Buckets

Despite my and my opponents' official ELO being ~620:

  • 53% of these games are marked from chess.com review played at a level above 1000

  • Only 21% were below 500

This suggests we are constantly matched against under-rated accounts.


8. Account Age
  • < 1 year old: 38%

  • 1–3 years: 32%

  • 3+ years: 30%

It's not just new smurfs. A large share of my opponents have well-established accounts.


9. Opponent Game Count

Average number of games played by my 47 opponents: 1,900

  • < 1,000 games: 55%

  • 1,000–3,000 games: 32%

  • > 3,000 games: 13%

  • Some had over 10,000, and one even 55,000 games

We're not playing against complete beginners.


10. "But You Don’t Castle..."

Every time I post an analysis, someone digs into my last games and says things like:

"You didn’t castle."
"You didn’t develop fast enough."
"You hung a piece on move 12."

Yes. Of course. I'm rated 600.

I try to follow principles: develop my pieces, control the center, look for checks, captures, and attacks. But I make mistakes. That’s part of this level.

My point is not that I play perfect chess. My point is that the games — mine and my opponents' — are being evaluated by the platform itself as high 800s or even 900+.

On these sample there are games rated 1300 or 1400 by Chess.com’s engine. Those ratings are capped by your actual ELO, meaning if you were higher rated, some games would be scored even higher.

So while people critique individual errors, they often ignore the bigger picture. The data shows that many games in the 600 pool are not being played like traditional 600-level chess.


💡 Conclusion

Chess.com’s 600 today is not what it was five years ago.

  • Many opponents have thousands of games

  • Precision is often above 70%

  • Games are cleaner and more disciplined

  • Over half of the games are played at a level above 1000

Climbing the ladder is harder when everyone is underrated. You might need to play like a 1200 to become 1000, because your pool includes a lot of hidden strength.

If you’re stuck, maybe it’s not just you. Maybe it’s that 600 today is actually 900.

Thoughts?

I've been stuck at 1000 for this whole year but I feel that that's just my current skill level

Avatar of Petrosian94
I_throw_in_the_Tal wrote:

Puzzle ratings and estimations don't reflect an "actual elo" or anything, but my estimations are usually within about 400 elo of my rating. And I suck. My games end when either myself or my opponent blunders big enough. I think that's pretty good for some nerdy get rich quick scheme that I have never paid for.

Also, kids absorb knowledge passively. Sometimes I forget how old I am. I couldn't reach 1000 elo in my lifetime even if my best friends were Magnus and Hikaru.

If you made an effort to play decent chess, you would climb above 1000 within months. Many of my students did.

Here are things I saw in your games:

1) you take little time

2) focus more on figuring out which squares are controlled by both sides

3) start taking into account what your opponenrs play

Avatar of DareTower
Yesterday I won a rapid game hitting 79,4 accuracy, my opponent had 63,2 accuracy. I got a 1050 game rating from it, opponent 600. Both players in the 300 category. So yeah the games can be tough even on the lower level. I just started with rapid games yesterday after I decided to take a break from the speedy games. I have played under 100 games of rapid in total. Climbing up in rapid isn't going to be easy. But I am focused on doing the lesson challenges. Those also help with puzzles because there is similar situations explained step by step. Not a lot of people mention the lessons yet those are very helpful. Even though I loved bullet I see that it can cause addiction and I don't want that so I am switching gears. I'm not saying that people shouldn't play bullet, I'm just personally quitting it (for now).