ChessBase statistics -- part three

8/13/2003 – A few decades ago a mathematician named Arpad Elo devised a pretty slick ranking system based on statistical probability. FIDE (the World Chess Federation) adopted it as an official standard in 1971 and chessplayers the world over have greeted each other with "What's your rating?" ever since. More in this week's ChessBase Workshop

ChessBase 14 Download ChessBase 14 Download

Everyone uses ChessBase, from the World Champion to the amateur next door. Start your personal success story with ChessBase 14 and enjoy your chess even more!


Along with the ChessBase 14 program you can access the Live Database of 8 million games, and receive three months of free ChesssBase Account Premium membership and all of our online apps! Have a look today!

More...

CHESSBASE STATISTICS -- PART THREE

by Steve Lopez

A few decades ago, a mathematician named Arpad Elo devised a pretty slick competitive ranking system based on statistical probability. FIDE (the World Chess Federation) adopted it as an official standard in 1971 and chessplayers the world over have greeted each other with "What's your rating?" ever since. (By the way, other sports have also adopted the Elo system, including FIFA, the World Soccer Federation, as a means of ranking national teams).

You might be wondering why we're starting this ChessBase Workshop with a history lesson. There are really two reasons. I occasionally get asked why a ChessBase search for [Capablanca/Alekhine/Morphy/Steinitz, etc.] coupled with an Elo search turns up no "hits" in the database. Since the Elo system wasn't adopted by FIDE until the early 1970's (in fact, there wasn't even a FIDE until the mid 1940's), pre-1970's searches involving player ratings will turn up empty. The second reason is that we're going to look at another player statistic feature of ChessBase 8 this week and, since it prominently features ratings, it's best if you use it on players active in the last thirty years.

I indicated in last week's issue that this feature is kind of "hidden". To get to it, you first double-click on a database (we'll use Mega Database 2003 in this article) and click on the "Players" tab. We'll have a look at Alexi Shirov's games as our example. Double-click on "S", then double-click on "Sh". Scroll down to Shirov's name, but this time just single-click on it to highlight it. Go up to the row of Toolbar buttons near the top of the screen and click on the "Statistics" button (which looks like a small blue, yellow, and red bar graph). After a few moments, you'll see this display appear:

This is a statistical display based on the games of Alexi Shirov.

The left-hand box shows his results sorted by specific players. It's a scrolling view, meaning that you can scroll up and down its length (using the scroll bar on the right of the box); you'll also need to scroll it from left to right (using the scroll bar at the bottom of the box) to see all of the information.

Let's look at the first line, just to get started. The first line refers to Shirov's games against Loek van Wely. Scroll the display to the right a bit to get "the rest of the story":

We can see that Shirov enjoys a lifetime record (as of Fall 2002) of earning 19.5 out of a possible 26 points in his games against Van Wely. The rightmost column shows the number of Shirov's wins: 13.

Note that you can sort this display by three different methods using the buttons below the box. The "N" button sorts the list by the number of games played. Clicking it now shows that Shirov has faced Vladimir Kramnik 60 times. Clicking the "A-Z" button sorts the list alphabetically by the opponents' last names. And clicking the "Result" button returns the display to the default view.

Here's a neat trick for this left-hand box: if you double-click on a player's name you'll get the ID Card for that player.

The middle box works the same way, but in this case the games are sorted by the event in which they were played. The first entry in this display is for a simul played in Andorra in 2001. Scrolling to the right, we see "15.5/20" and "+11", meaning that Shirov scored 15.5 out of a possible 20 points and won 11 games outright.

There are buttons below this display, too. "Result" and "A-Z" work exactly as described above for the left-hand box. "Date" sorts the events in chronological order, while "Table" will generate a list of the opponents in that event and the result against each. Double-clicking on the name of an event in this list does the same thing as clicking the "Table" button.

Moving to the right-hand box, we get a list of ECO codes and some percentages:

This gives statistical information on how well (or poorly) the player has done in various openings. The first entry for Shirov is for the A90 ECO code. The left-hand percentage shows his performance as White, while the right-hand one shows how he did as Black. So we see that Shirov wins 45.60% of his games as White in this opening and 16.50% when playing the Black pieces.

Note that it's perfectly natural for the Black percentage to be much lower than the White figure. It's an old chess adage that you play for a win as White and for a draw as Black; top-level players typically follow this, which explains the disparity in the numbers.

Under this box there are three buttons that allow you to sort the list. "White" is the default display option, sorting the openings according to the White percentage (from best to worst). "Black" sorts the list according to a player's success with the Black pieces, while "A-Z" sorts the list by ECO code, in alphanumeric order.

There's one more button in this display that bears mentioning (and takes us back to the "history lesson" at the start of this article). If you click the "Elo" button at the lower left, you get a whopping huge bar graph which looks like this:

If you want to follow along, you'll need to try this in CB8 (which you really should have been doing anyway -- wink, wink); I had to reduce the size of the display to get it to fit in most web browsers without making you scroll the browser display). The lower graph shows the number of games Shirov played with his opponents divided by Elo ratings. The numbers along the bottom of the graph refer to Elo ratings while the numbers along the left side of the graph refer to the number of games. Thus the higher the bar, the more games played against players in that Elo range. We see that the majority of Shirov's opponents fall into the 2600-2650 range with 328 total games played.

The upper graph gives performance ratings for Shirov against his opponents. The numbers along the bottom of the graph again refer to rating ranges, while the numbers along the left refer to the Shirov's performance rating. Here's an example. Let's look again at the 2600-2650 range (in which Shirov played 328 games against opponents in this range). CB8 takes the opponent's ratings for those 328 games and figures a performance rating based on the ratings of the opponents in the 328 games and how well Shirov did against them, coming up with the figure 2704. Shirov performed at a 2704 level in those 328 games.

A text line across the top of this display gives a pile of useful information about the player. "Score=" refers to the player's overall performance. "Score White=" and "Score Black=" gives figures for how well the player did when playing a particular color. The "+" figure shows the total number of wins, the "-" figure shows total losses, and the "=" figure displays the total number of draws. The "Elo" figure shows the average rating of all of his opponents, while "Perf=" shows the player's overall performance rating.

So what does this mean to the average player? It's simple -- if you keep a database of your own games, you can use this statistical feature to evaluate your performance. You can find the "holes" in your opening repertoire as well as make determinations as to how well you do against players in different rating categories. And that's some very useful stuff for evaluating your own chess play.

Until next week, have fun!



© 2003, Steven A. Lopez. All rights reserved.


Topics cb8
Discussion and Feedback Join the public discussion or submit your feedback to the editors


Discuss

Rules for reader comments

 
 

Not registered yet? Register