### Thursday, March 24, 2005

## Book Review: _Chess Exam and Training Guide_ by IM Igor Khmelnitsky

This is a great book. It presents a hundred problems of all sorts--tactics, positional play, openings, endgames, everything-- and it's very good just as a problem book, although the price would be steep for a 200-question puzzle book. (100 diagrams, 2 questions per diagram.)

But it's not just a puzzle book--it contains a complicated scoring key, where you enter your results for each problem, and it scores you in a dozen categories based on your answers (every problem counts for several categories).

Here's an example (from the introduction):

White to play.

1. Evaluate the position A. White is winning B. White is significantly better C. White is slightly better D. Nearly Equal |
2. Where should the White King go? A. 1. Kd6 B. 1. Kf6 C. 1. Kf4 D. No significant difference. |

Stop here if you want to work out the answer. Coach K recommends taking up to 20 minutes on each problem.

OK, the answers are 1. A and 2. B. Those are worth 5 points. Also, 1. B. is worth 1 point; but 1. D. is so bad you lose a point. (You can have negative scores; this is part of a strategy to make guessing unprofitable.) According to Khmelnitsky, 60% of players below 1000 rating got part 1 correct; 75% of players from 1000 to 1800; 80% of A players and Experts; and all the masters. On part 2, 20% of novices were right; 29% of 1000 to 1400-rated players; 45% of the class C and B players; 62% of A players and Experts; 82% of 2200-2400 players, and all the players above 2400. Khlemnitsky also gives breakdowns for each of the wrong answers.

This problem would score in four categories: Endgame, Attack, Tactics, and Sacrifice. After the probloems are tables where you take your percentage score in each category and find your rating for that category. Now, doing ratings estimation in a puzzle book is nothing very new--Nunn does it, Emms does it, everybody's doing it--and I'm sure that trying to break the results down by position type has been done somewhere as well. The difference is that Khmelnitsky has tried the problems out on a variety of players of varying strengths over the past few years, so when he says that 82% of masters got this problem right, but less than half the players below class A, he's got _data_.

Now, I have no way of knowing for sure the quality of his data, but I've never seen a problem book give an estimate anywhere near my actual rating. _Chess Exam_, however, told me that I'm 2200. My current rating is in the 2150s (both USCF and FIDE) and my peak before my hiatus was about 2250, so I'd call that pretty much on the nose.Between the detailed answers to the puzzles, the carefully-described methodology, and the very accurate overall estimate, I've got a great deal of confidence in K's other estimates. And in the final part of the book, Coah K gives advice: what books to study and what methods to use, for each categry, based on your score.

So you get a puzzle book, a rating estimator, an assessment of your weaknesses, and an improvment program, all for $24.95. That's a great value. Anyone from Class E to Master will get a lot out of this book, and there are not a lot of chess books you can say that about.

Highly recommended.