Chess.com ratings are so inaccurate

It's your own fault you don't know how to address these attacks. I offer coaching services to help players deal with time and space warping matrix agents.

thanks but I don't have no money for that, besides you might be affiliated with the matrix, these topic are always full of matrix agents

I looked at your profile, and your last game and I can say, yes you are a beginner. You need to study how to play instead of ranting in the forums.

I looked at your profile, saw your ratings and I can say, yes you are a beginner. You need to study how to play instead of bashing others in the forums.

1500 is intermediate

1500 especially on chess com is beginner

It's 'bashing someone' to state the obvious about a sub 400 player claiming he's not a beginner. A sub 400 who claims the site is rigged and people like you are all robots. I don't go ranting and raving calling people robots or claiming there is a system in place to keep me where I am. I actually have learned how to play. If basketstorm is a friend of yours, you do him no favours by enabling his paranoid delusions either about the state of the website or his playing ability. 

Then what are you, who is under 400. sub-beginner?

Oh very easy these are not really my games.

No definitely not he gives like 1800+ level chess advice when he makes mistakes

I would also another reason that ratings don't stabilize is rating deflation; lets say two players, who are actual 1200 players play a game. One wins, or maybe they tie, but the sum of their scores stays the same. However, each of them get just a bit better. This effect is obviously tiny, but tiny effects multiplied by a huge number of games across a large number of players add up over time. This causes ratings to never settle, because this deflation would mean that the actual playing skill that defines 1200 (or any number) even in the pool of the same players, is NOT constant. That being said, I am not able to collect data to test this deflation hypothesis. I think the data I collected already is pretty strong in convincing that ratings do not converge no matter how many games you play, but I think it helps to try pin down the (likely multiple) sources for the lack of convergence

And also if they both are underrated they will stay underrated. And when they beat a player of same rating who is not underrated, they will surely move up but they might also face an even more underrated player after that and their rating will go back. Tight rating-based matchmaking does all that + lot of existing inaccuracies. People with same rating but varying skill (stronger or weaker than you) come and go and your rating is like on waves, it does not depend on your skill it depends on how accurate ratings of others are. Like if you are 400 and many new stronger than you players (like 800-strength) will register as 400, your rating will sink and their rating will stay around 400 because they will mostly compete between themselves without playing with actual 800s.

I'll repeat my points:

- initially players must not be rated and they must not affect ratings of others. See how it's done in FIDE for example

- matchmaking must be forcibly wider (this won't be unfair, you will lose only a little from your rating if a higher-rated player beats you and later you will beat a lower-rated and gain those few points back), doesn't mean you won't get equally rated opponents, you will, but accidentally

- pool-joining matches and recalculation of isolated pools must happen constantly, to align game outcomes with rating differences. FIDE did such alignment this year

There's certainly a lot more volatility in play at the lower rated levels, but that's because there's a lot more volatility in the play: they may be a good player overall but keep making mistakes because they're not ocncentrating or careless: they may play a good move without understanding why it's good. People can improve significantly with just a few tips, whereas when you get to higher ratings, people make fewer mistakes and tend not to improve without making a serious effort. The problem is not the rating system but people - we're not machines, we have good days and bad days.

Agree plus rigging happens all the time, consistency is not possible

And it will happen no matter what rating system is used, including the one you've proposed

narrow (or really, any non-random match orchestration where there is a correlation between the ratings of the two players) could cause pool separation between groups of ratings, which could lead to the probabilities like I have experienced.

I feel like you can call any rating range "beginner"......

Just like in the 2024 FIDE tournament the other players coulda called gukesh a "beginner" because he was rated lower than almost all of them, doesn't change the results of the tournament.
I fail to see what that has to do with discovering constructive ways to make ratings more meaningful.

As for classifying basketstorm's games, looking at them he seems to mostly play fast games (in which moving faster can matter alot) and just blunders frequently. I do see some things that I do different strategy-wise that I do not see in basketstorm's games (planning castling, for instance), so idk I'm no chess teacher.

So, finally got some time to play consistently-ish (took more than 2 weeks though), restricted my opponent rating range and..... watched my rating continue to climb....... it climbed to 1260 when I was playing opponents around 1200 (not an unpredictable result innate of intself), so I didn't change it for a bit, rating stayed around 1260ish, raised opponent rating range again, watched my rating climb again right past 1300, then past 1400. Appears to follow what I had said about the rating settling point/ converging a different value depending on the ratings of opponents because the win probabilities really aren't what they should be. Whether this is the case across alot of profiles or just mine could yield some insight as to deficiencies in the rating system, this type of analysis would need to be done on a large number of profiles. I'd rather actually impact my rating by playing better rather than using math, but it is hard to argue with results.

More variation in lower ratings could mathematically mean that games against them should affect ratings less like fitting weighted least squares regression. calibrating this to counter the possible rating deflation I mentioned earlier may be a way to improve rating calculations