Saturday, April 24, 2010

Evaluating Positional Evaluation

I promised Andres Hortillosa I’d give a review of his book, Improve Your Chess at Any Age. I have several books of this theme, Rolf Wetzel’s book and Michael De La Maza are a couple on the “top shelf” in this theme. I am going to be brutally honest and keep this as respectful as possible. What I liked was the fact that this was a guy with a similar back ground as I in the “work” world. He has an engineering background with an eye for process controls. So I definitely liked the idea of seeing what he had to say. The problem I had with the book was that there was no clear instruction of how he applied a “six sigma” method to improvement. Instead, he approaches some key themes, gives some sage advice in the form of platitudes but then dives right into “ here are my examples”.

He does give a formula for a thought process ( slightly paraphrased):

Step 1: Initiate a broad reconnaissance of the position to gather key data elements
Step 2: Search for Specific threats
Step 3: Rank the severity of threats
Step 4: focus your response against the highest degree threat
Step 5: search for Candidate moves
Step 6: Execute the move in your head
Step 7: Conduct a post –reconnaissance in your head as if the move was made
Step 8: If a problem with 7, go back to step 5.

The key to step one is properly evaluating the position. You can read Silman, Nimzovitch, or any other Positional manuscript but inaccurately evaluating the position over the board is a show stopper. I see it in my games. Steps 2-4 again are critical areas where I need to improve. I like Heisman’s approach to looking for “ checks captures and threats”. But if you inaccurately evaluate a position in the first place, you may miss things making the rest of the steps difficult.

There is talk about his approach to doing tactics which made sense. In short, he feels its more important to understanding the setup before the light comes on and says “ Mate in such and such”. I have to agree.

Going back to my desire to apply a six sigma approach to improvement, without going into great depths of statistical methods and cpK analysis of process control charts, I decided to do an initial evaluation of my games this year focusing on my improper positional evaluation.

I need measurable metrics I can tweak that can give me some indication that the trend in my rating is within better limits instead of thrashing about. I said it before, consistency in my playing strength is what is preventing me from reach the next class level. So, I shall set this as a goal to attain.

So where are my process control charts? Here, I agree with Mr. Hortillosa. Achieving a rating on a tactical server or program is not a good indication of improving my process over the board. It’s merely an indication of my ability to solve puzzles. Mind you, there is some merit behind the puzzle solving but since I have not seen a major jump since my third circle of hell ala De La Maza’s method, I am approaching this a lot differently.

First and foremost, I need to fix my ability to evaluate my own positions. Like any process that needs fixing, first it must be assessed. I am painstakingly going through each of my most recent games and looking at the graph of positional evaluation my chess engine produces.

Let me walk you through an example of a game I played in March against an 1800 player. Here is the positional evaluation graph. It looks like Fenway park’s green monster.

NN1 V Blunder

Move 13 begins the green monster but a subtle positional error occurred on move 11.

It is Black’s move 11 (mine), I was incorrectly looking at provoking White’s d-pawn at the risk of weakening Black’s queenside and advanced to c6-c5. Thinking out loud, White’s position is freer though slightly behind developing the Queen bishop. Black’s pieces are out but the light squares are cramping the Black’s Queen Bishop. The computer points to a game where 11…h6 was played and gives Black a better game. ( 11… Rad8 was another suggestion that leads to a draw… but more on that a little later).

The position speaks to giving the light squared Bishop a little escape to keep pressure on the b1-h7 diagonal. Interesting how I was fixated on opening up the d-file. What benefit was I anticipating from this? Opening up the d-file before placing the heavy artillery behind the pieces makes no sense. I jumped the gun on this. Opening up the position gives White a strong center with a knight on d4. ( I didn’t steps 2-4 any justice)

In hindsight, weighing the positional requirements versus creating a tactical thrashing is a lesson I can walk away with. In this sense, Black’s development gives him a marginal plus but that is in contrast with White’s more open position as it is a matter of time before the Q-bishop is developed. More important for Black is to improve the position with either more room for the light squared bishop or building up the forces on the d-file ( like Rad8) BEFORE attempting to open it up.

After White exchanges 12. cxd5 Nxd5 13Nfd4 Black makes another positional gaff, 13…Bd3 shown here:

This is a weak tactical double threat against white’s rook on f1 and c4 pawn. I failed to look at White’s counter threats ( again not even considering steps 2-4) when I calculated this. White followed up with 14. Nb5 attacking my queen and Bishop and proceed to play into an exchange of 2 pieces for the rook. I underestimated ( ignored or fell blind to) the threat that Nb5 had on the position. The lesson here is to make sure I follow through evaluating my opponent’s checks captures and threats on every move. I don’t do this enough.


LinuxGuy said...

Great post!

That's a tough position. ...e5 looked good. After ...c5, giving White the minority attack Black plays ...a6 and then White should play Re1 to stop ...Be4, then Black can play ...Rfd8 (basically to protect the bishop on d6 - White likely plays Qf3) and then ..Rac8.

Definitely visualize the move before playing it, that is a biggie and will help particularly in time-pressure.

Look for pins, forks and checks since that is what tactics basically are.

Another thing I've learned is don't get caught up in notating the moves. A lot of the 'A' players I just played against were already doing what I had recently learned do, which is not to write down the other guy's last move until you are prepared to make your own; sometimes minutes later they write down my move. This way, notating won't distract your ability to make the right play on the board.

LinuxGuy said...

I forgot to say also look for strong retreats.

Does your opponent have a quiet move that shuts down your planned attack? A piece can move backwards and then be double-attacking. It's normal human psychology to not consider strong retreating moves, IMO.

In that same vain, don't play something just because it's forcing. People as well as computers have a habit of going for the most forcing moves. There may be a better positional way/solution at hand.

castlerook said...

Thanks for sharing this!

Have you read Jonathan Rowson's Chess for Zebras? It's my all-time favorite chess book (actually one of my favorite books--period), and it has a lot to say about the psychology of playing one's best each game, as well as the challenges of adult chess improvement.

You won't find much in the way of clear, step-by-step methods for improvement, but I honestly don't think you need such a thing.

Anonymous said...

In my experience, positional errors can make my game more uncomfortable or turn an advantageous position into a disadvantageous one, but I loose almost exclusively due to tactical errors worth 1 pawn or more. At least that's how it is on my level as a B player and this also holds true in the games against people up to 2000 and a bit beyond that, it's just that they make less tactical mistakes and play much better positionally. But I couldn't point out a single game against anyone <2000 that I lost because of positional mistakes alone.

LinuxGuy said...

I've found the same thing to be true as Anonymous said, except they do make positional errors, the only thing is that you still have to win the game against them once the error has been made.

You need the time, the energy, the skill, the motivation, be more on your game than they are on theirs (sharp), and even the belief to win the game after that point. Did I mention nerves of steel?

Liquid Egg Product said...

On the plus side, Andres Hortillosa completely nailed the game show host/car salesman look.