Smyslov_Fan wrote on 07/03/11 at 03:13:05:
Uhohspaghettio wrote on 07/02/11 at 17:43:59:
Smyslov_Fan wrote on 07/02/11 at 17:40:09:
Uhohspaghettio wrote on 07/02/11 at 17:33:37:
...
I don't know why GMs play the King's Gambit a lot more than the Latvian Gambit. The King's Gambit scores only about 45% in comparison to the King's Knight opening which scores about 55%. I can't find a large enough sample of Latvian Gambit games by high rated players but it would seem to have to score 35% or less to be worse than the King's Gambit (since Black has an expected score of 45% by playing 2. ...Nc6).
Wow.
I wonder where you get your statistics from. In the last decade in games between players rated +2400, the King's Gambit scores ~55% according to my database. Historically, in all games in my database, it scores ~57%.
Are you sure you're looking at the right stats? I think you may only be looking at White's wins, not the overall percent.
The fact that you don't know why GMs play the King's Gambit while avoiding the Latvian may be a clue for you to try to figure that out. This may be the key to your understanding of the relative virtues of the two openings.
I was looking at it yesterday on Mega Database, also you can see from this:
http://www.365chess.com/opening.php?m=3&n=5&ms=e4.e5&ns=3.5 (
switch to master's database) Black has the advantage.
The 365chess.com "Master's database" only has 542 games in the King's Gambit. It doesn't say whether "masters" were playing both sides or not. It's easy to cherrypick games when the sample size is small. Try comparing GM vs GM games.
I trust chessbase databases far more than that one. Please try running your Megabase with games played only by players rated +2500 (on both sides).
Please, look at your site again. It shows 2.f4 scores 49.1 wins for white, 18.5% draws and 32.4% wins for black. That is after 11,854 games.
That comes to about 58.35% overall winning percent for white.
After 1.e4 e5 2.Nf3 scores ~56.35% according to the same database. It is also played far more often. The online database that you mention has 255,305 games.
The Latvian only has 1521 games in that database. Even though it's played as a surprise opening, White still scores 56% according to that database.
The numbers you're referring to are of the complete database, not just master games. I never take the whole database into consideration because you could have people who don't know what they're doing or are falling into the same traps that a good player never would. I wasn't cherry picking, I never take the complete database.
I tried Megadatabase from 2000-2010 with both players over 2500 and it has just 90 games and there it does have a percentage of 51.7% for White, while 2. Nf3 has 55.6%.
However if you bring it down to both players over 2300, there are 557 games and White's score is only 45.5%, which is similar to what's given on chess365.com for masters' percentage and what I gave yesterday. (When there's a lot of games I have to right click on the moves for the percentages to come up). 2. Nf3 gets 55.0%.
I don't know why the KG seems to do better for players over 2500 while also better for the whole database on chess365.com but much worse for over 2300. It could be that these players weren't playing the KG in quite as serious games, mainly ones they knew they would probably win easily or ones that didn't matter. 90 games is probably just not enough. If for example a really top player took it up frequently, then it would skew the result significantly.
Smyslov_Fan wrote on 07/03/11 at 03:13:05:
Of course, statistics alone don't describe the relative viability of an opening. But when even Stefan Buecker acknowledges that the Latvian gives white a significant edge and the statistics are so overwhelmingly in favor of white, perhaps it's time to re-evaluate.
I didn't argue that the Latvian Gambit was a good opening, just that it was potentially viable and arguably at least as "good" as the King's Gambit.