I think that the level of risk taken by one of the world's top players is usually closely related to a) tournament situation b) level/rating of opponent c) clock times/time control d) psychological factors.

I see your point Smyslov_Fan, but it seems you're limiting risk-taking to the player's subjective experience at the board, i.e. Shirov didn't feel he was risking much against Smeets at Corus (I assume that's the Petroff game you mentioned) because he miscalculated and didn't see the danger he was in.

But there is also objective risk. When GMs enter complicated, tactical positions where mistakes are more likely to happen, most of them are humble enough to admit that at any moment they may miss some complicated line that Rybka or Fritz will point out to any patzer following the game. And if you're allowing a winning tactic that Rybka would spot, there's a chance that an on-form GM opponent will also spot it. 

So objectively, Shirov was taking a risk against Smeets, and Anand was taking a risk against Shirov when he allowed the winning ...Ng3! Because even GMs miscalculate and make mistakes, entering any kind of complicated middlegame or endgame involves risk compared to just playing out a theoretical draw, and they take that into account when preparing for a game. But for players like Shirov and Topalov going for complicated positions is still the practical decision (the one that maximizes expected value, in the language of economics), because that is where they are strongest compared to their peers. They realize fully that they will likely lose more games than a Kramnik or a Leko, but when on form they will also win more. It's a calculated risk, not a reckless gamble.

And that's what I believe Avni was onto: GMs will make decisions that risk losing the game when that is still the best practical chance of maximising the score. The overall result is less risk-taking than your average hacker. But unless they're taking short draws every game they can never play without any risk at all.

SWJediknight wrote on 02/05/10 at 13:43:45:


Maybe I should have just said that GMs maximise their chances - not necessarily of winning every game, but of scoring as many points as they realistically can given the level of opposition and their own strengths and weaknesses. If a GM plays in opens a high percentage of wins is expected, less so in a super-tournament where even the tournament winner often has more draws than wins.

I've always had the impression that many (though certainly not all) GMs play primarily not to lose, rather than primarily to win, because getting a mix of wins and draws gives better results than a mix of wins and losses- similar to the "practical" philosophy in football of using a 4-5-1 formation, playing with the draw in hand and hoping for a 1-0 (take Jose Mourinho's Chelsea a few years back for example).  Many articles in Chess Monthly in particular have given this impression, with Peter Leko particularly noted for such a tendency.

That said, I'd certainly agree that Kasparov usually played to maximise his chances of winning rather than playing not to lose, and there are other examples.

I also get the impression that the higher your level of play the less likely it is that risk-taking will work, and the more likely it is that encouraging the opponent to take risks will be beneficial.

One of the conclusions drawn by Avni in "The Grandmaster's Mind", where he interviewed and tested Israeli top players, was that GMs are ruthlessly practical in their approach, compared to average players. They play to maximise their chance of scoring the full point, not to gamble and realize dreams of wild slugfests and brilliancies. 

Tony's attitude looks like a good example of this.

Of course some GMs still have a marked preference for tactical and attacking play (i.e. Tal, Shirov, Topalov), but that makes sense if that really is their greatest strength compared to other GMs. On the other hand many club players who think of themselves as "attacking" players are really not that good at calculation, but they don't seem to connect the dots.

I know several people (typical tournament players) who play very solid with White (e.g. English, Catalan) but play very aggressively with Black (e.g. Najdorf). Their reasoning is that "with White you can win by doing nothing, but you need to take chances if you want to score well with Black."

msiipola wrote on 01/30/10 at 10:12:47:

In such circumstances, I think it is better to give your opponent the opportunity of taking himself a risk. It feels a lot better when you "force" your opponent into a sacrifice where he seems to get promising attacking chances and you actually get very good couterplay