About    Using this site    Historical ratings    Formulas    Source data    Future projects    Contact
   Welcome to the Chessmetrics site   
March 26th, 2005: Hello, I'm Jeff Sonas and I'd like to welcome you to my new and improved Chessmetrics site. This website is devoted to statistics about chess. Since the summer of 1999, I have spent countless hours analyzing chess statistics, inventing formulas and other analysis techniques, and calculating historical ratings. This website allows you to explore chess history "by the numbers" in an interactive way. You won't find any analysis of chess moves here, but you will find historical ratings and many other statistics that can't be found anywhere else in the world. In addition to estimating the chess-playing strength of individual players throughout chess history, I have also invented new ways to rate the strongest tournaments and matches of all time, as well as the best single-event individual performances. You can find lots of colorful graphs showing the rating progression of top players throughout time, and also age-aligned graphs, so you can see who were the most successful players at various ages.
In late 2001 I released my original Chessmetrics site with my prior effort at generating historical ratings. I have made considerable improvements in the quality of my data, the clarity of my rating formula itself, and also in the way that I present the information to you on the website. Nevertheless, for now you can still visit the old Chessmetrics site.
If you'd like to jump right in and start exploring, I encourage you to go to the Summary tab and start clicking around. There are lots of links that will open up new sections with additional data. Or, if you would like to read more of what I've written about the historical ratings, the formulas, the underlying data, or the future projects I am envisioning, you can click on one of the links below the logo. Or if you want to go through things in order, click on the link below to read more about how to use the site. If you have questions, comments, suggestions, or corrections, feel free to send me an email. Please use this website responsibly, and enjoy yourself!
Read about how to use this site
   Some frequently-asked questions (and their answers):
Question #1: How do I find a particular player's historical ratings?   Answer: Just click here to search for a player by name. For future reference, you can always just click on the "Players" tab, and then click on the "Find a Player" link.
Question #2: How can a player have a peak rating higher than their best performance rating?  Isn't that impossible?   Answer: No, it's not impossible. It would indeed be impossible with traditional performance ratings, but not with Chessmetrics performance ratings. A Chessmetrics performance rating indicates the overall rating we would assign to the player if we only knew about that one event, and nothing else. Because an individual event typically covers only a few games, the performance rating calculations for single events are fairly conservative; we don't want to award extremely high ratings based upon just a few games.
For example, although a 50% score in a 6-game match against a 2800-rated player would traditionally be called a 2800 performance rating, my formulas (which are sensitive to the number of games played) would conservatively call that a 2728 performance rating, because of the small number of games. If that single six-game match were the only results we had available for you, we would assign you an overall rating of 2728, not 2800. Hopefully it makes sense that a high percentage score across many games would have a greater impact on your rating than if it were across just a few games. It's more evidence that you really are that strong. For instance, if you had scored 50% in a 24-game match (rather than a 6-game match) against the 2800-rated opponent, you would get a 2795 Chessmetrics performance rating.
Thus if you played a total of four different six-game matches against that same 2800-rated opponent, with the same 50% score every time, your individual event performance ratings would each be 2728 (as we have already seen), but your overall rating, based upon all 24 games, would be 2795. Thus it is quite possible to have an overall peak rating that is larger than any of your individual performances, simply because several events, viewed as a whole, constitute more overall evidence that you really are that strong, compared to the evidence from just one event. For more information, you can click here to read about how the formulas work.