Do bots have an accurate rating?

Hey guys,

I started to play chess seriously like some weeks ago. I dont feel like I am improving or maybe I am wrong. I didnt try yet to play against players, my actual rating is 500 i think. But I am trying to beat all the computer bots. I already managed to win some 1800, 2000 bots but there are some 1300 bots that I just cant win, they dont do 1 mistake, always defend perfectly. Are those bots accurate, rating wise? To be honest I dont know If I am not improving or some of those 1300 bots are actually 2500. Thank you !!

I am 1700-1800 in 10 minute time controls and I lost a fair amount of games against Antonio (1500) with 10 minute time controls.

I am playing without time to practice more and have more time to think. I managed to win some 1800 bots, but there are some 1300, 1400 i just cant win. Dont understand that.

The answer is that the bots have no rating at all. That is just a number that some chess.com programmer came up with (by guessing), and that's it. It does not go up and down as they win and lose games. If the original estimate was wrong, it will stay wrong forever and no one will ever bother to change it.

The original bots ratings seem to be based on the rated games they used to play in live chess.

They are very UNDER RATED

I'm 900 and I can beat a 1400 bot so no the bots rating is not very accurate at all

I beat Noam (2200) ONCE and I also beat a fide master in 10 minute chess ONCE but it doesn't mean their ratings are not accurate.

Ohh, ok thank you all. Its wierd as I sometimes feel like the 1300 bot outplay me in all the ways.. I do a play like thinking 2...3 moves ahead and the bot already acts like he knows what I am doing 100% of the time. No mistakes.. its rare to see it making a mistake. Weird..its said as would be so great if their rating was accurate.

If you don't feel like reading, skip to the end.

Hello! I'm a bit late. I played a game against the 'Deadlost' bot (1300 according to the game) and when I saw the post game 'rating from one game', it said it was a 650. Now, the system isn't that accurate (I made some awful moves and got a 1050, while they were just barely behind me), and this is from one game, but it seems that the bots ratings don't always correlate with their play. It's like they'll play great and then make 3-4 really terrible moves that most 1300's wouldn't make. This is pretty much what happened in my game. They edged me out in the Opening (87.1 to 86.4) and the Middlegame (83.5 to 83.2) but DIED in the Endgame (76.8 to 17.8). The game was 72 moves long and pretty much over when they hung their Tower (1 of their 2 pieces that wasn't a pawn or a king) to my knight on move 27, but still.

Point is, the 1300 bots play like 1300's-2500's like 60-80% of the time but play like 300's-700's the rest of the time. Hope this helps a little.

From my experience… no.

Seriously, I’m like, 850, and have consistently beaten Antonio (granted, with a few takebacks) even though he’s 1500. The bot ratings are wack.

im 1700, yet I can beat Danya (2650) so, probably not

I'm only 1000 and can beat a 1600. Bots are overrated. Literally. The ratings are inaccurate.

I float between 900 and 1030 and I have never beaten a bot higher than 1700. The 1200 and 1400 bots make some pretty dumb moves at times and miss some obvious mistakes that I have made.

No entiendo

I just hit 1200 and can beat 1800 bots sometimes, can beat 1500 and 1600 more frequently. The bots are programed to blunder at certain rates, they might already be to far ahead when they blunder or may be you don't know it is a blunder at the time it happens.

I've played lots of bots and 5 in person USCF tournaments. I have rating of 1290. It appears the bots make good, excellent, and book moves more frequently as their rating goes up but the randomization of mistakes is nowhere near the distribution of errors expected from a human. I just beat Julia (1800) who just dropped a bishop! I have a 3 game winning streak against 1800 bots and have never beaten an 1800 human, so bot ratings are just rough guides of relative strength.

No. I have beaten an 1800 bot. I have never beaten a class B player (1600-1800). Ever.

No, I don't think its accurate, as I'm rated 750 rapid and was able to beat a 2200 rated bot. I ran a couple of tests and the reason for this appears be that the way they play just has to result in a single game based rating estimate close to what chesscom says the bots rating is. The problem is, it dosen't matter how it acheives such an estimate, so where a real 2200 player would play nice moves and make a mistake or two, it's possible for the bots to play the best move on the board 30 moves in a row and then blunder mate in one, as long as that results in an estimate close to the rating chesscom gave the bot. And that's exactly how I beat it. I was able to hold the bot off long enough (I lost 2 bishops in the process) and then it blundered mate in 4, and my tiny little brain managed to see it.