bobobbob - try chess tactics.
Every tactic is rated and solving each one (or not solving) changes your rating - so you would basicly know your "rating" before playing lots of games.
This is still only one of many factors - opening knowledge, endgame skill, strategy & positional skill, etc.
Reading Silman's Amateurs Mind is a perfect book that illustrates how someone can be a higher rating than another, but still have horrible knowledge in many areas.
There is definitely no correlation between % of blunders and ratings. Blunders are often the result of calculating several moves a head and getting an idee fixe. A high rated player is more likely to make blunders than a low rated player looking one move at a time. Rubinstein one of the all time great players was very prone to making blunders.
That's what happens when you talk about stuff that you don't have a clue about.
I agree. I think Kaufman put out some numbers about gross blunder ratio to rating too. Forgot what they were, but it was something like every 200 points is 4x less likely than the last, starting at 1400 and giving it in terms of # of moves. Something like that anyway... but it was there :)
Fritz has a function "rated games" where you can set its "Elo strength" and play it, and after a few games it will give you an "Elo rating".
However I think Fritz's ratings are too high. In Fritz rated games I currently have Elo 1400, while OTB I'm somewhere between 1200-1300 (not officially rated yet but I see by my game results so far). It's also quite easier for me to beat Fritz with "Elo 1540" (that's the lowest value it allows me to set) than a human 1500 player from my club.
I would rather have a chess programs that based on my games suggest that I study
opening or tactics or endgame or look at tactics number 34,55,66 or something like that.
Has anybody tried www.ChessZen.com Looks like their ELO estimate by analyzing the games is accurate. They also have estimated ELO ratings of various world champions including Morphy, Fischer and Kasparov.
Reading this back - eighteen months have passed - it seems I was merely poking fun at this statement made by the poster above me. The sentence probably looks this strange due to some editing mistake... or because he really has a high rating and did not look ahead one move at a time.
lol
Entertained much, are you?
As an IT guy, with a science and programming background, I see NO reason whatsoever to think that an algorythm could not be developed to be able to analyze a game and assign an 'instantaneous ELO' at each move.
If you blundered, your iELO would drop, perhaps dramatically. Also, the first few moves as was pointed out above would have a very high iELO but that would drop once the players got out of theory. Unless they were Magnus.
As an IT guy, with a science and programming background, I see NO reason whatsoever to think that an algorythm could not be developed to be able to analyze a game and assign an 'instantaneous ELO' at each move.
If you blundered, your iELO would drop, perhaps dramatically. Also, the first few moves as was pointed out above would have a very high iELO but that would drop once the players got out of theory. Unless they were Magnus.
I don't trust it. I may pick sub-optimal moves from the engines POV because of a plan I am following that is clearer to me than the engines choice would be.
For example, if it is an endgame and I am up on material, I may give up some material to simplify, reasoning that it reduces my chances of an error. Since the engine doesn't make the kind of error I am afraid of, it will give the move a thumbs down. But I maintain my results will be better in the long run doing it this way.
Different chess engine programs have automated players of different strengths you can play against.
It's a good estimate. If you can beat the 1800 computer "personality" easily, but you have a hard time with the 2000 personality, you can probably safely say you're around 1900-2000.
I agree that it wouldn't be perfect, but it would continue to learn with every game. Although the number of possible games is a Very Large Number, the number of probable games is much lower. Within 5 years I bet a quantum computer would be able to give a very accurate iELO.
I disagree with the idea tha tthe end result is all that matters. Sure that is how ELO is calculated, but that doesn't mean its the only way to measure a game. 1-0 might be replaced by a graph showing iELO throughout a game. I just see ELO as a simplistic system that doesn't really capture all aspects of the game. I might be 1400 right now, but I've played 1800 games.
I believe the performance of a single game can be determined to within a reasonable comparison of an agreed upon rating system (USCF, FIDE).
For example, in Fisher's rematch with Spassky, the playing strength was estimated at 2600+.
If a new chess player loses to 10 straight GM's his rating would not be a good estimation of his strength. If you looked at the games and realized that he always hung his queen within the first few moves of each game, you might conclude his playiing strength is around 600.
There must be a way to get a rating from a game. There are many different potential moves with each turn and chess engines can analyse how good each one is fairly well. It stands to reason that during the course of a game a computer could work out how good your choices were and compare that information to a database to get a rating level. I understand that one game wouldn't be very accurate but a larger database must give a good indication. It should also be able to identify your best games.
I like" analyse your game app" as I can see trend of my game as compared with my opponent and the level of play for that particular move as measured by ELO. It gives elo rating compared to the engines best choices fre the particular move. It can be induced that you are being compared with a highly rated play and the elo can be an indirect measure of how well you are playing and which part of your game needs more fixing. There are graphs and other tools to further see the intricasies of the game. I am not saying that it is perfect but it definitely is practical and useful.
I always love to check my play with this app before I actually start analysing my game. I want to hear of others' experience.I have also other tools such as droidfish, anlyse this etc.
Recently I am playing on Lichess but I can not directly import my games to analyse you game app. What I do is I share it to analyse this and then save it and retrieve it with analyse your game." Analyse your game" does not work well with Lichess.
As an IT guy, with a science and programming background, I see NO reason whatsoever to think that an algorythm could not be developed to be able to analyze a game and assign an 'instantaneous ELO' at each move.
If you blundered, your iELO would drop, perhaps dramatically. Also, the first few moves as was pointed out above would have a very high iELO but that would drop once the players got out of theory. Unless they were Magnus.
This is basically what Gothamchess does live on Twitch from time to time (also available on YouTube) and he's pretty consistently right about the players' ratings on chess.com. You do need knowledge of a specific pool of players to make an educated guess about the rating from a single game.
A neural network trained on a hefty sample size of games from e.g. chess.com could probably give a reasonable estimate of your rating in that specific pool of players.
1. Nc3 Nh6 2. Nb1 Ng4 3. Nh3 Nf6 4. Ng1 Ng8 1/2-1/2
Estimated rating result:
White: 3200
Black: 3400
The chance of a game with only Knights is like 0,00000000000000000000000000000000000000000000001%
Is there a program that can analyze a database of your games and then approximate your rating?
This topic is 11 years old, but yes, there is an online website that can calculate your rating based on your games:
https://www.chess.com/play/online
It is obvious that ratings change over time based on the changing population that the ratings are based. Therefore any calulation of ratings would be an estimate. But it would be useful and interesting to have an objective measure of chess strength based on the moves in a game alone.
It would be interesting to know just how good Karl Marx was. Or Napoleon.
The problem with this kind of thing is that ratings have no absolute meaning, they just show where you stand within the body of players that is being rated. A given rating at chess.com represents a different level of skill that a USCF rating, which in turn might represent a slightly different level of skill than a FIDE rating. Basically, the larger the pool of rated players the more meaningful that rating is, as there are more likely to be truely high level players. If I started my own chess league but we only had a couple of hundred players and only one master player that one master would soon have a super-grandmaster rating if we were holding a large number of events, as he would be head and shoulders above everone else in the ratings pool and always win, but he might have a skill hundreds of points weaker compaired to someone who played in international FIDE events with the same rating number.
So in short, attempts to do what the OP is talking about are no more than guess work and chicanery.