Evaluating Elo rating on a per move basis

Question

I've been a Chess.com user for a little over 3 years now. I play EVERY day, usually have anywhere from 10 to 50 games going on any single day.

Since I've started, my rating has only gone up about 150 points. For the past 6 months, my Elo has stayed fairly close to 1400.

I know Chess.com has the feature of evaluating the games that I've played and suggests what I did wrong, but I have to request all of those games to be reviewed.

Is there another tool/app that can evaluate the strength of individual moves during each and every game, that would tell me the "strength" of the move? For example, if I stupidly put my queen in a position where the opponent can take it, then that's a blunder and would be rated very poorly.

Or, is there a way to automate all of the Chess.com games to be reviewed by the engine to determine how I did in each?

Answer 1

Is there another tool/app that can evaluate the strength of individual moves during each and every game, that would tell me the "strength" of the move?

To be able to evaluate the strength of individual moves, one needs to estimate the difficulty level of the position. This estimation rests on two ingredients: the ability to evaluate the position (almost) perfectly, and a sample of responses for that position from a pool of players with different strength. This is thus an empirical question.

For tactical training, there are straightforward implementations, such as this one:

http://chesstempo.com/user-guide/en/tacticRatingSystem.html

You can add to the accuracy of a solution a time element:

http://chess.emrald.net/time.php

For any kind of position, I believe it's still an open problem.

---

A related question is the optimality of play. To that effect, Guid & Bratko did an interesting study:

The basis for their evaluation was the difference between the position values resulting from the moves played by the human chess player and the moves chosen as best by the chess program Crafty. They compared the average number of errors in the player's game. Opening moves were excluded, in an attempt to negate the progress in chess opening theory.

http://en.wikipedia.org/wiki/Comparison_of_top_chess_players_throughout_history

But such study cannot estimate the strength of a player as requested by the querent. It can only evaluate optimality of play, and order players according to this evaluation. This does not give ELO or complexity of any given move.

Answer 2

Yes, such tools exist and they are called chess engines. Look at stockfish, houdini and rybka there are a lot of GUIs (graphical user interfaces for them). They will report inaccuracies, mistakes and blunders and will show you better lines.

Nonetheless you will be able to automate this process, but the question is 'what will you do' when the engine will analyse your ~ 33.000 games? To actually get something from this analysis - you need to go step by step through a game and make some thinking. So I would rather forget about all those games (may be selecting a few that you think are super cool) and start analyse new games that you will play.

Also irrelevant to the question, but as suggested by others learning chess and playing chess is completely different (also might sound the same). I have seen many players who play tens of games per day (mostly 1-3 minutes, where half of the moves are completely random). They randomly win/draw/lose, but stay on the same rating, because they do not learn from their mistakes.

From my own experience (which basically means nothing) I was staying at the same 1400 rating for 4 month randomly playing games. I also thought that this might help, but it did not. Then I started to analyse my games and played longer games (10 minutes), play puzzles and learn openings. I played at maximum 4 games per day and in 6 months I was able to boost it to 1800 and improve it even further later. Take a look at my answer to see some insights how you can advance your game.