Mateo Help

Sort:
Avatar of PANVidarr
I primarily play against bots, mostly because of the infinite time limit (I have kids so I can’t always focus for a full 6-minute Blitz game).

I haven’t played enough to know my ELO. It takes hundreds of games for it to settle and I just haven’t played that many. Not to mention they changed the “starting ELO” from 1200 to 400, so the ELO game is basically nonsense now — most games I do play with humans I’m paired against a 450 but they make zero blunders, see every tactic, and use complex gambits. It seems to me that by lowering the starting ELO all they did was slowly shift the numbers down so that the median player is now considered 400.

Based on chess puzzles, performance against bots, game review, and every other online trick for predicting ELO, I believe I’m somewhere between 1000 and 1400:

* If there’s an immediate fork, skewer, or pin I almost never miss it — but I don’t see 2 moves into the future
* I sometimes accidentally make a move that lets the enemy fork me, but never straight-up leave a piece undefended
* I tend to always “win” the opening (Stockfish shows +0.5 to +1.5 in my favor when we start mid game)
* I know how to take advantage of pinned pieces and overloaded defenders
* I’ve memorized a handful of end games (pawn and rook, queen checkmate, two bishop checkmate, ladder mate, etc)

With all of this, I have a 75%+ win rate against every bot up to Mateo in challenge mode.

But against Mateo my challenge win rate is 1 win to 20-ish losses. I’ve beaten him in friendly mode with take-backs (try a move, 3 rounds later realize I’m in a bad spot, back up and try something else) but I just can’t figure him out otherwise.

What’s frustrating is no matter how much I improve or what my “post game rating” shows, Mateo is always one step better. At first I would get 1000-1150 post game rating and he would have 1200-1350. Recently I get 1800-2000 post game rating and he gets 2100-2300.

Is there a trick to this bot? Some tactic he’s programmed to miss or blunder he’s programmed to make? Because it feels like he’s just always doing best moves and there’s never an opportunity to punish mistakes with him like I can on other bots.
Avatar of PANVidarr
Wow, the iOS app silently strips all formatting and white space and truncates my post… brilliant. I couldn’t figure out how to share a link to an example game from the app. But I generally make no blunders and still lose.
Avatar of PANVidarr
What’s frustrating is that he’s always “one step better”.

When I first started playing against Mateo I’d get 1000-1150 post game rating and he’d get 1200-1350. Now I consistently get 1600-1800 and occasionally 2000, but he wins with 1850-2200.
Avatar of PANVidarr
Is there some tactic he’s programmed to fall for or something? Some trick I’m missing that makes this bot possible? Some strategy I’m not aware of?
Avatar of HeckinSprout
PANVidarr wrote:
I haven’t played enough to know my ELO.

Yes, you have played enough to establish your rating. According to your profile page you have played 48 rapid games against human opponents. Your rapid rating is very clearly 474, with a peak rating of 573. If you are playing bots to judge your rating, you should know that the chess.com bots are severely over rated. They are good for someone's self esteem, but not for judging your chess strength.

You mentioned your puzzle rating. Puzzle rating doesn't correlate to chess rating. It's a different system.

Lastly let's talk about the game review feature. It's not a good source of truth. As an example, you could play a game and the review would tell you that you played like a 1000. I could play the exact same game move for move, and since my rating is 1350, it would tell me I played like a 2000. It's like that to give you a boost to your ego.

I'm not saying any of this to discourage your chess journey, but to be realistic about it. It's easy to beat a bunch of higher rated bots, and then get really upset at yourself when you can't beat actual players in that rating range. But the good news is you can work towards it. And there's lots of people and resources here that can help you.

Lastly... Mateo. I just finished playing him. It's been ages since I've last tried. https://www.chess.com/analysis/game/computer/616046237/review I didn't play anything special against him. He hung a pawn on move 3 and brought his queen out very early which I bullied to gain tempo. Advanced tactics really aren't needed to beat him. Get your pieces developed, castle. Connect your rooks and get them on open files. It's mostly just good chess principles and not blundering.

Avatar of PANVidarr
You should really stop perpetuating the lie that ELO is well-known after a 2-digit number of games. I have written dissertations on the use of ELO in games and I can assure you that even if those 48 games were all played by me, and I weren’t improving between games, that’s still not enough for your ELO to be “settled”. In theory once you reach your “proper” ELO you should have a 50% chance to win any game. But if you’re currently ranked below your current ELO this doesn’t imply a 100% win rate (bad days, unfamiliar gambits, etc). I won’t bother with all the math but to know your ELO plus or minus 200 points requires about 60 games and to know your ELO plus or minus 100 points requires closer to 120 games. That said: I share my account with my 6 year old son, so half my games are him playing — and four weeks ago I didn’t know the first thing about chess, so I’ve been rapidly learning. The last 5 times I actually played a human myself I won 5/5, so I’m pretty confident I’m above 500 ELO. I’m not just relying on bit games and post-game review to make claims. I’m comparing the moves I’d made to watching official tournaments, I’m reading books which propose puzzles to measure ELO, and I’m considering my personal win rate (not chess.com account win rate). I am quite confident in my estimation of 1000-1400.
Avatar of HeckinSprout
PANVidarr wrote:
You should really stop perpetuating the lie that ELO is well-known after a 2-digit number of games. I have written dissertations on the use of ELO in games and I can assure you that even if those 48 games were all played by me, and I weren’t improving between games, that’s still not enough for your ELO to be “settled”. In theory once you reach your “proper” ELO you should have a 50% chance to win any game. But if you’re currently ranked below your current ELO this doesn’t imply a 100% win rate (bad days, unfamiliar gambits, etc). I won’t bother with all the math but to know your ELO plus or minus 200 points requires about 60 games and to know your ELO plus or minus 100 points requires closer to 120 games. That said: I share my account with my 6 year old son, so half my games are him playing — and four weeks ago I didn’t know the first thing about chess, so I’ve been rapidly learning. The last 5 times I actually played a human myself I won 5/5, so I’m pretty confident I’m above 500 ELO. I’m not just relying on bit games and post-game review to make claims. I’m comparing the moves I’d made to watching official tournaments, I’m reading books which propose puzzles to measure ELO, and I’m considering my personal win rate (not chess.com account win rate). I am quite confident in my estimation of 1000-1400.

Jeez, I try to be nice and get my head bit off. A few things:
1. Chess.com does not use "elo". It uses Glicko.
2. Elo is not an abbreviation or acronym so you don't capitalize it.
3. If you really think your rating is 1000-1400, challenge some players in that rating range to a game or two.
4. Sharing your account is against chess.com's TOS. A six year old playing chess here is also against the TOS. Minimum age is 13. They created Chesskid for younger kids.
5. You obviously are not interested in advice so I will disappear into the ether now.