I guess it would lead to an attempt of dissecting what amounts to "greatness" in chess. Doesn't it seem oversimplified that Elo rating is the one and only measure? I agree it gives you a tool "This is how good you are!", but it does not tell you "This is why you are so good!", or "This is how you could get even better!" Say, you are a developing chess player, but appear to be stuck at say, 2000. How would you know what area to focus on to get better if you do not know what constitutes a succesful player. Say, you are not blessed with a guru, an all-knowing master, who sees through your blatant weakness and bammm, comes up with a fix and there you go to 2400.
Moneychess?
Sort:
I would like to see which players have the highest winning% as white, and the lowest losing% as black. Whining% would be an important factor in assessing a player's potential I would imagine.
Funnily enough, you can look at drawing percent as a marker for a strong player. It takes a fair amount of technique and knowledge to draw a good portion of your games.
Good point. It still bugs me when a player simplifies into an objectively drawn position against a certain player, as part of a strategy in a tournament. Or goes for a perpetual when there's more to be had, because a draw is all that's needed. But of course many draws are hard fought and exciting.
The reason a more nuanced metric system would be helpful, because it would help to judge the areas that would need improvement. On one hand one can look at players playing certain openings as their chance of success is obviously higher with the most frequently played opening. One can look at number of moves played till decision or draw. One can look at draw achieved after how many moves. One can break down opening move numbers, middle game move numbers and endgame move numbers.
But it would also help to find a system of self help: visualization, calculation, mate pattern recognition, tactical awareness, strategical planning, etc.
What a profound question you have asked. My chess buddy and I spend most of our time together on such quandaries over how one measures up. He is at 1,700, and I'm 850ish (blunder problems - so consider the source). Yet my problem solving (stepping away from the board) is at his level. Putting aside ones goals (other than the obvious, of winning and increasing your rating (I offer Michael Lewis's book "Playing to Win" for perspective on your question). In which he "connects the dots" as to why there are so many tournament draws as a means to satisfy the point system. ELO is a game changer! As a believer that improvements are specific and the outcome is not always linear, one also needs to be in touch with the whole input/output expectations. I often wonder whether time spent playing, is time lost studying (the psychology of learning).
Hey if anyone could help me I would really appreciate it. I'm trying to find the best way to export data for bullet games with specific opening filters.
I'm really struggling just with that part but after that, I want to try to use the vast amounts of data to predict human behavior based on moves rather than the computer suggested moves to look for:
- Sort data by rating range
- Moves that generally prompt a longer response for an opponent
- Moves that generally prompt suboptimal play from opponent
- Moves that are frequently (successfully) followed by a series of moves that can be played regardless of the opponent's response
- Basically anything else like this that I can come up with
Any suggestions?
Dumbluck626 wrote:
Hey if anyone could help me I would really appreciate it. I'm trying to find the best way to export data for bullet games with specific opening filters.
I'm really struggling just with that part but after that, I want to try to use the vast amounts of data to predict human behavior based on moves rather than the computer suggested moves to look for:
- Sort data by rating range
- Moves that generally prompt a longer response for an opponent
- Moves that generally prompt suboptimal play from opponent
- Moves that are frequently (successfully) followed by a series of moves that can be played regardless of the opponent's response
- Basically anything else like this that I can come up with
Any suggestions?
You would have to find members that play bullet and manually download games to do anything like that.
Watching the movie "Moneyball" one wonders if there could be a "sabermetric" equivalent in chess. Are there any statistical measures, yet to be discovered quantifiable skills that would rank and characterize players other than the Elo rating? Say, winning percentage with white, with black, draw percentage, win percentage against higher rated opponents, loss percentage against lower rated opponent, etc. Any ideas or suggestions?