Saturday, March 31, 2007

Analysis improvement tips from the masters

Once you've selected some candidate moves, it is time to analyze them, to think through variations and decide which candidate is best. In a previous post I described what the masters have to say about when and how much analysis is needed.

What most of us want to know, however, is how to get better at analysis. Herein I include three different techniques recommended by three different authors. If anybody has other suggestions, please let us know!

A. Kotov: learn to analyze by analyzing
Kotov's suggestion, in Think like a grandmaster, is a natural one. To get better at something practice doing it!

Specifically, Kotov suggests picking a good annotated game collection and working through the moves in a game until the position becomes fairly complicated. He then puts the book away and "set myself the task of thinking long and hard so as to analyse all the possible variations. [...] I would sometimes write down the variations I had examined and then I would compare them with those of the annotator. At first there was a big discrepancy in favour of the latter, but then I leraned how to widen my scope and delineate each variation with considerable exactitude."

I am using this method, as it most closely resembles what I have to do in real games. I write down all the variations (I know this detracts somewhat from the "real game" goal, but it helps me to compare my analysis to the annotations).

B. Soltis: read through master games
Soltis' method is somewhat similar. He says:
A worthwhile exercise is to play over an annotated game [...] Stop at each point where there is a short comment, one that gives an alternative variation that runs one or two moves into the future. Once you visualize that note to its end, continue playing through the game until you rech the next short note and repeat the process. When you've gone through the entire game and exhausted all the short notes, go over the game again and try to visualize slightly longer notes. [...] Eventually work yourself up to the five-movers. But ignore the really long notes (unless you are a masochist).
This is a nice method which I have used a little bit. One shortcoming compared with the Kotov method is that in real games you don't just have to visualize, but visualize moves that spring from your own imagination.

C. Buckley: memorize the squares and visualize auras
Buckley, in his popular book Practical chess analysis recommends Kotov's method but recommends starting by memorizing the board. He also says that it is helpful during analysis to visualize lines of force emanating from the pieces, lines that pass through other material on the board, as this helps you avoid making blunders such as not correctly visualizing that a file has been opened by one of the variations.

I know a lot of blindfold experts say it is absolutely essential to have the board memorized, and I even posted about various software available for this skill. This technique seems more important if you are interested in blindfold chess. I have just begun visualizing lines of force (this is something my coach suggested as well), so it is too early to judge how helpful it is.

9 Comments:

Blogger takchess said...

I forgot the name of it but I am sure others will know it. There is a methodology where one studies a position for an hour or so and writes down all the plans one sees. It is rather involved and I imagine helpful.


I read about it in an Alburt book .

3/31/2007 08:52:00 PM  
Blogger takchess said...

upon reading your blog entry again, it sounds like the Kotov suggestion.

4/01/2007 09:01:00 AM  
Blogger Blue Devil Knight said...

Tak: yes, that is what Kotov suggests. On its face, it it is the most reasonable way to train on analysis. I've been doing it about a week now (using positions from Euwe's 'Master vs Amateur') and it is teaching me a lot about what kinds of mistakes I make (e.g., 'always look at forcing moves' is a slogan we all know but I sometimes still neglect).

Also, especially important is the danger of thinking that because I spend a lot of time analyzing a position, after a couple of moves I can select a move based on that previous analysis. Always dangerous. Ya' gotta look at the present position as it is easy to overlook things when working through game trees in the mind.

4/01/2007 10:07:00 AM  
Blogger Pale Morning Dun - Errant Knight de la Maza said...

Silman made the same suggestion as Kotov, or perhaps he got the idea from him. I have not tried in practice however. He recommends having a little notebook set aside for the task. He says at first things will be chaotic and lengthy. Eventually, your notes will be more streamlined. Of course, your success has a lot to do with what games you pick out too I think.

4/01/2007 04:35:00 PM  
Blogger Blue Devil Knight said...

PMD: as long as you use Fritz or another trusted advisor, you should at least get a good assessment of the relative values of various moves. I have been doing the Kotov method with positions from my own games (the positions where the evaluation function goes against me most drastically).

4/01/2007 05:41:00 PM  
Blogger BlunderProne said...

I think its important to practice how to analyze games as already stated by everyone. Its part of De groot's observations of masters vs amateurs. While training for tactics helps in the pattern recognition, practicing to improve your analysis accuracy is the other part of his equation.

Several methods are out there as you suggested. Whether you use, the square's method, Kotov's method or Silman's imbalance method ( or other variations) the important thing is to find a method that suits you and STICK to it.

I practice a combination of Silman's and Kotov's by doing three things. (1)I go over my own gmaes and annotate them. (2) I go over a collection of games. I go between Zurich 1953 by Bronstein and the Games collection from GM-RAM to pick and choose. Usually on a weekend I set up the board, go through the first few opening moves then take my time at critical points and jot down all that I think is relevant about that position. I try to simulate an OTB tournament condition while doing this except to have a notepad as a hard copy of my thought process. I found hte first few times I did this, I was jotting down EVERYTHING and it was very ineffecient. I found though that I can quickly get to the essence of the position now and spend time evaluating realistic candidate moves trying to place a value on which one is better.

The third thing I do as an exercise is to do solataire chess from the CL column. I score consistently in teh 1800's ( I wish my rating reflected the same).

4/01/2007 06:46:00 PM  
Blogger Pale Morning Dun - Errant Knight de la Maza said...

I'd like to add onto what Blunderprone said about Solitaire chess. Pandolfini has a collection of the games in book form, and I just realized that one technique that is simple and straight forward is to go to the archives and pull the game up, then hide the notations window, and play the game in solitaire fashion. This makes it easier as far as keeping the scores and answers hidden. Not all the games are in my megadatabase, but many of them are. Its another way to review master games. Further, these games are all chosen for their high tactical content so it's good practice from that standpoint.

4/02/2007 05:34:00 PM  
Blogger takchess said...

I have done the solitaire chess with Master vs Amateur which is good because you have a clear explanation of Euwes thought process to his move selection. Also the name I was searching for earlier was Stoyko Exercises.

4/02/2007 06:25:00 PM  
Blogger Blue Devil Knight said...

Since my goal is to do one game an evening, I don't do the solitaire thing for a full game. I pick a single position to study in depth.

What I did was to look through all the games, find positions that look complicated and for which Euwe gave a lot of alternate variations with explanations, and wrote down that move on the first page of the game (and the side that I am to analyze). Now that I did that, I just pull up the game for the night in Fritz (the pgn is online), play up to that move rather quickly, and then do analysis for about 20 minutes (I should probably do about 40, perhaps I'll build up to that). Once I write out all the variations, I then go back and read through the book, paying special attention of course to that move, and through the rest of the game.

I am trying to do this every other night, and on alternative nights I play a slow game (40 30).

4/02/2007 11:08:00 PM  

Post a Comment

<< Home