New player beating advanced bots, what’s my ELO?

Hi everyone,

I'm a new player who has been exploring chess.com for the past few weeks. While I've only played a few rated games against real opponents, and my rating is still stabilizing, I have some experience with the site's bots. I've managed to defeat all the bots rated up to 1600 ELO. Although the more advanced bots present a tougher challenge and I lose more often than I win, I can still beat them with focused effort.

Given this experience, I'm curious about what my actual ELO rating might be. It seems like the ELO ratings of the bots don't quite match up with human players of the same rating. Has anyone else noticed this discrepancy? What ELO range should I expect to be in based on my ability to beat these bots?

Thanks for your insights!

I think the bots between 1400-1600 are about 80% of their ratings, while those under 1400 feel like they are 50% of their ratings.

After you play one of the bots, check the review and see rating what they played the game at.

You must have started with an ELO that was way too low. I have trouble beating 1600 and 1700 bots consistently, but my rapid rating is double yours. If you want to play a 24 hour "daily" game, let me know.

I've recently been working my way up through the bots - only moving up in rating if I beat them four times in a row, twice as black and twice as white. I've beaten a 1400 and will be moving up to a 1500 bot next (although I have beaten 1600 bots in single games). My chess.com rating is just over 700 though, and actually I think that overstates where I am - I rarely beat a 700 person so I'm probably more accurately in the 600's. Thus, I suspect the bots are over-rated by maybe 1000.

I can beat all the bots up to 1600 challenge mode

I like that idea of not moving up until 4 consecutive wins. Maybe I'll end my current tilt working on that.

Your rapid rating is 483, so that’s your “elo”.

Rating in review means nothing and will change when your rating changes. Just like with brilliant move. Don't believe in that stuff. Tricky bots like Wumpus could be picking skill level randomly at start of the game. And in general, because bots here run locally in your browser or on your smartphone, and we don't know if their thinking is time limited, using chess.com bots as measure of your skill won't be accurate. Use bots from some Desktop software like LucasChess and don't set the time limit for bot's thinking.

1600 bot like 800 human

the bots are all slightly overated, but great job! probably around 600 - 900

Playing a 600 bot is definitely easier than a 500 live player. I currently practice and usually win with the 1000-1300 and there is no chance I could beat a real player that high. Zero. But the 3200 bot seems ridiculous. What's it like to play that for strong players? I just get destroyed so fast. Even with take-backs, it just re-adjusts in 2 moves and destroys me in another way. I at least got to an end game with Eugene.

600 bot like 200-300 human, 1000-1300 bot like 500-600 human, 2000 bot like 1000 human.
but high than 2500, It's almost real chess.com blitz rating.

Bots normally correspond to FIDE ratings of late 1990s, early 2000s. Also they can be calibrated to a different time controls, that adds some variance, +/- 150 maybe.

Chess.com players' ratings are very deflated on lower end compared even to modern FIDE.

But also chess.com bots are sus, several users reported that on weak devices bots play weaker. It's confirmed that chess.com bots run using YOUR local resources (smartphone or PC if it's browser).

Chess.com uses Komodo. You can run Komodo bots locally. Komodo itself doesn't provide Elo setting like Stokfish does. There's only skill setting and guidance: Blitz time control, 0 skill is 250 Elo, with each new skill point add 125, for two last points add 200 and 200. Again this Elo is related to FIDE Elo of late 1990s-early 2000s.

I wanted to revive this old thread because I think the conventional wisdom on bots is missing some key ideas, although some people in this thread hinted at them.

The chief difference between a bot and a human, is rated games, when you lose one, you lose elo. When you lose against the bot, most people don't count it. They remember wins and not losses, because this is human nature, without the elo score to keep you honest, human frail memory takes over. That's a huge difference

Let's say they say they beat a 1500 bot, but they actually aren't that good. They remember the times they win, they discount the times they lose. Don't consider yourself as able to beat a 1500 bot unless you play it 10 games and win 5 of them.

I call this concept "bot elo" and when I first tried it, I was shocked to find I was much worse against the bots than I considered myself. I found I was only 1400 "bot elo"( meaning in a round of 10 games, I would win at least 50 percent of those games) , much to my disappointment, and then gradually worked myself up to 1600 "bot elo". I'm still a noob, so that is where I stand today. Now I can occasionally beat a 2000 even a 2200 bot, but only once in a blue moon.

The second thing people don't do, is they don't click on the almost hidden "custom" setting and set the time control for the bot. It's way harder to beat a bot in 10 minutes, then to have unlimited time - you set up the time pressure, and you suddenly find your bot elo isn't the "1600" you just claimed.

The claims that bots don't play like humans and so on and so forth are not exactly important. yes, if you have traps you set, a human might fall for them. Example given, you castle, trying to trigger your opponent to castle, but you actually set up an attack for that position. Bot isn't psychological like a human, I agree. But it isn't that big a deal for most people, the reality is a noob has weak vision of the board, weak tactics, weak tempo to the game, poor end game, and so on and so fortha nd a bot a some level is goign to give you the smack down until you correct your weakness. If you aren't getting smacked down, play a harder bot at that point.

I think bot practice gets a bad rap because of how the interface encourages wrong results. Giving you a crown for defeating a bot once is absurd. Not having time controls by default is absurd. Chesscom can play these bots against humans and get real elos for them. That's another miss. If chesscom did this for those that volunteer, and refund elo to people afterwards - they have a huge audience and this wouldn't be any kind of burden to get the true elo and many would volunteer. Frankly anyone could do it if they wanted to research the topic, I am not that interested, not important. They are inflated, we all know that and it doesn't matter, they work great.