kwchess
|
|
« on: Feb 17, 2010, 08:07:47 PM » |
|
Would anyone here know of the statistical occurance rates of ending types, or where I could find/generate them?
If we were able to find out the most commonly occuring endings, would it be possible to then create custom problem sets that occured based on the true occurance percentage with a premium membership?
|
|
« Last Edit: Feb 17, 2010, 08:27:00 PM by kwchess »
|
Logged
|
|
|
|
kwchess
|
|
« Reply #1 on: Feb 17, 2010, 08:20:18 PM » |
|
Aha:
15.91 rook rook 12.73 rook & bishop rook & knight 6.50 two rooks two rooks 6.35 rook & bishop rook & bishop (same color) 6.20 bishop knight 5.82 rook & knight rook & knight 5.40 king & pawns king (and pawns) 3.62 rook & bishop rook & bishop (opposite color) 3.52 queen queen 3.33 rook & bishop rook 3.11 bishop bishop (same color) 2.94 knight knight 2.84 rook bishop 2.67 rook & knight rook 2.09 bishop bishop (opposite color) 1.90 bishop pawns 1.83 rook knight 1.73 knight pawns 1.69 queen & minor piece queen 1.53 rook two minor pieces 1.41 rook pawns 1.30 queen rook & minor piece 1.26 rook & pawn rook 1.05 rook & two pawns rook 0.79 queen pawns 0.75 queen rook 0.58 queen two rooks 0.43 king & one pawn king 0.32 queen minor piece 0.17 queen & one pawn queen 0.15 queen two minor pieces 0.04 bishop & knight king 0.02 queen three minor pieces
The original data (Müller, Karsten; Lamprecht, Frank (2001), Fundamental Chess Endings, Gambit Publications, ISBN 1-901983-53-6) had percentages were games that went into those particular endings, rather then the percentage of endings. I've normalised them to 100% so that it now shows the percent occurance of a particular ending if the game gets to that stage.
Some of those are drawn, so they could be removed and the data re-normalised. It's nice to know, as someone new to ending theory, that by learning to deal with 5 of the most common endings that you will be able to apply it in ~50% games that get to that stage (note even taking into account the fact that you can deliberately force the game towards endings you know well).
|
|
« Last Edit: Feb 17, 2010, 08:39:47 PM by kwchess »
|
Logged
|
|
|
|
kwchess
|
|
« Reply #2 on: Feb 17, 2010, 08:24:15 PM » |
|
So now that we know roughly what to expect, would it be possible to set up endings with the premium membership so that the RvR, RBvRN, RRvRR etc.. occur with the percentages shown above?
I think this could be a valuable feature, focus your time on the endings that you are actually likely to see!
Any comments on this are welcome of course, in particular I would be interested to hear what the stronger players think, given that my perspective on the chess world is somewhat limited.
|
|
« Last Edit: Feb 17, 2010, 08:34:42 PM by kwchess »
|
Logged
|
|
|
|
kwchess
|
|
« Reply #3 on: Feb 17, 2010, 09:26:09 PM » |
|
Thinking about this again....will the problems actually already appear at approximately these percentages since they came from real games?
|
|
|
Logged
|
|
|
|
richard
|
|
« Reply #4 on: Feb 18, 2010, 12:03:14 AM » |
|
Hi kwchess,
I find the muller data somewhat confusing as it is not clear if some of those endgames also have pawns attached, wikipedia's article suggests they do, but then the data also lists separate endgame types with pawns. I suspect they do, because I've seen other frequency data that has KRPKR as the most frequently occurring endgame and about 3-4 times more common than KRKR, so I suspect to get "rook rook" this high would require the inclusion of the KRPKR endgames.
Currently CT doesn't have weighted endgame distribution as an option, although the current distribution does somewhat reflect the reality of what is found in games, it is not going to be a perfect match for several reasons: - The CT endgame generator prefers positions people find difficult (e.g. where the player in the game played a drawing move where a winning move available). - The CT endgame generator applies its own weighting, for example it under samples KRPKR otherwise these are so common that they would have overwhelmed the set. It also under samples (or completely removes) endgame types that are trivially won, so for example KQRRK and other endgames with overwhelming material advantage are not particularly interesting (apart from avoiding stalemate perhaps). - Certain endgame types tend to cluster around rating ranges, so while the entire set is somewhat similar to real game frequency, the problems in the rating range you are currently working at may not be.
Richard.
|
|
|
Logged
|
|
|
|
drahacikfm
|
|
« Reply #5 on: Feb 18, 2010, 10:09:46 AM » |
|
kwchess, also note that currently the Chess Tempo endgames only have 5 total items on the board. So you can't get KRP versus KRP, or KRBP versus KRNP, or any double-rook endings, etc.
I don't think it's particularly important to be able to make a personal problem set with those exact percentages. A better approach is to just look at the percentages to see which endgame type is most common (rook and pawn endings) and make a set to practice those. Then when you have had enough of that, make a set for the next most common endgame and practice that, etc.
|
|
|
Logged
|
FIDE Master Drahacik
|
|
|
|