chess statistics: with what opening does white win most?

[flok]

Hi,

Just for fun I analyzed aprox. 1 milion chess matches (that did not end in a tie) to see which openings win most. The result can be found below. I ignored the openings that were played less then 1% of those 1 million. I'll list the 10 with the biggest number of wins for white and the 10 lowest number of wins for white. The value f is number of white wins divided by number of black wins and n is the number of times this opening was used.

Code: Select all

+------+--------+-------+
| eco  | f      | n     |
+------+--------+-------+
| D36  | 2.0794 | 13826 | 
| C10  | 2.0636 | 14697 | 
| C41  | 1.6970 | 21363 | 
| B09  | 1.5733 | 15130 | 
| C06  | 1.5329 | 14741 | 
| B31  | 1.5081 | 12083 | 
| D35  | 1.4946 | 14020 | 
| E11  | 1.4857 | 12384 | 
| C42  | 1.4738 | 23622 | 
| D85  | 1.4080 | 11938 | 
+------+--------+-------+
10 rows in set (4.12 sec)

Code: Select all

+------+--------+-------+
| eco  | f      | n     |
+------+--------+-------+
| B20  | 0.8256 | 17137 | 
| B50  | 0.8324 | 19940 | 
| B25  | 0.8405 | 14064 | 
| A48  | 0.8932 | 19400 | 
| C01  | 0.9037 | 26337 | 
| B21  | 0.9426 | 20070 | 
| B02  | 0.9429 | 15478 | 
| A03  | 0.9452 | 14912 | 
| A46  | 0.9619 | 16619 | 
| A04  | 0.9779 | 15785 | 
+------+--------+-------+
10 rows in set (4.16 sec)

This and more of these kind of chess statistics can be found here:
http://www.vanheusden.com/cchess/datami ... chess.html

[Ovyron]

Interesting, where is E00? I think it's the best try from black to avoid D36 from the get-go.

[flok]

Code: Select all

+------+--------+------+
| eco  | f      | n    |
+------+--------+------+
| E00  | 1.1462 | 3410 | 
+------+--------+------+

but as you see that's only with 3410 matches while I analyzed 1 milion, that's not significant I guess?

1.1462 is less than the factor that white wins more often than black in general which is 1.2215 for 4 milion games played

[Ovyron]

I liked to see how reliable it is in principle. It's more of a "Black, avoid this variation!" than a "White, play this variation!". Seems a case of "with what opening black loses the most."

[TL]

Would it be possible to count moves that go forward or backward in a game. For example, can you determine how often or what percentage games have 3 moves in a row that go backward. ..10.Ne4-e2 Ne5-f7 11.Qd4-d2.. ?

TL

[flok]

Would it be possible to count moves that go forward or backward in a game. For example, can you determine how often or what percentage games have 3 moves in a row that go backward. ..10.Ne4-e2 Ne5-f7 11.Qd4-d2.. ?

Tomorrow I'll write a script which will calculate this for you.
I expect the results somewhere sunday: got a nice fast underclocked pc (q9400 cpu) but 4 million games take a while to analyze

[flok]

Would it be possible to count moves that go forward or backward in a game. For example, can you determine how often or what percentage games have 3 moves in a row that go backward. ..10.Ne4-e2 Ne5-f7 11.Qd4-d2.. ?

Tomorrow I'll write a script which will calculate this for you.
I expect the results somewhere sunday: got a nice fast underclocked pc (q9400 cpu) but 4 million games take a while to analyze

A little sooner than expected but the results are:
games: 4258385
moves: 329020187 (white + black)
white 1 move back: 19744495
white 2 moves back: 3206322
white 3 moves back: 544140
black 1 move back: 19320550
black 2 moves back: 3174928
black 3 moves back: 537528

Please note that the counts are cumulative: if there were 3 moves in a row that moved back, it is also counted as 2 and 1 move.

[TL]

Thanks flok. It looks like you put the moves together in one big string then went thru them. Do you have this program available on your web site or is it available to the public? I'd be interested in looking at other move sequences.

Tl

[flok]

Thanks flok. It looks like you put the moves together in one big string then went thru them. Do you have this program available on your web site or is it available to the public? I'd be interested in looking at other move sequences.

Not yet but it will. Can't say when.

But before then you're welcome to get me to do other analysis!