Author: Robert Hyatt
Date: 09:48:35 09/17/99
Go up one level in this thread
On September 17, 1999 at 11:44:06, Helcio Alexandre Pacheco wrote:
>I've studied this chess theme carefully enough to get a master view of it (IMHO
>:).
>There are many factors that you should consider in your evaluation of the
>position.
>First of all: I put pawn majorities in 3 class: good ones, bad(or weak) ones and
>unclear ones.
>An example:
>a) White Pawns on: a3,b4,c4,f4,g3,h2 Black pawns: a7,b6,e6,f7,g6,h7
>In this case, White has the good and black has the bad one.
unless maybe both castled queenside? then black's majority turns into a
distant passed pawn.. Although I assume you are factoring in distance to
queening?
>b) White Pawns on: a2,b2,c2,f3,g2,h2 Black pawns: a6,b5,e5,f5,g6,h7
>In this one, White has the bad one and black the good one.
This is one I might not agree with. IE if the kings are at g1/g8, I would
probably rather be white, although it wouldn't be clearly better for either
as the white passer is farther away and requires that black's king immediately
take off for the queenside to stop it, while the white king can try to
infiltrate and win one or more black pawns...
tough one..
>c) White Pawns on: a2,b2,c2,f2,g2,h2 Black Pawns: a7,b7,e7,f7,g7,h7
>In this position, it's unclear if the majorities are good or bad.
Here I personally factor in king positions. With the pawns not advanced
at all, it probably is a non-issue as you suggest.
>
>One important thing here is that the side with the good one have a great control
>of space and the center and therefore will have the better king in the endgame
>(in the first position white can put his king on d4 easily... in the second,
>black can put his on d5...).
But can either keep them there? If I push the pawns and make a passed pawn
on the queenside, your king is definitely going to have to go over to stop it.
Hence the decoy ability of a distant passer to win the game. But there are a
lot of tempo issues to be sure.
A more common case is white with pawns on a2/b2/c2, f2/g2/h2, black with pawns
on a7/b7 and pawns on f7/f6/g7/h7. Trading into this position with black is
_really_ asking for trouble, and if you move the kingside white pawns a bit
(double two of them as mentioned earlier) then white gets a passer, black
can't, and black probably loses.
>
>From what I read, It seems that your code already do this classification.
>
>Then we have to consider the other pieces in game. As you said, the advantage of
>a majority is bigger in the endgame. But, in which endgames will it be good
>enough to give an decisive advantage? Depending on the pieces, a bad majority
>can become a good one!
>
>Studing some master games I was able to identify some favorable endings where
>the advantage of a pawn majority is (nearly) enough to win. For example:
>
>1) Rook endings: The advantage is not so decisive. There are many rook endings
>where one side is a pawn down and can still draw. So you see the position more
>or less as a pawn down endgame where the side with the bad majority has good
>chances of holding. The side with the good majority can keep good winning
>chances if he have the active rook and can cut enemy king.
>
>2) Double Rook endings: In this kind of endgame, other themes are usually more
>important as the control of the 7th, active vs passive rooks. But again, good
>drawing chances.
>
>3) Queen endings: The side with the majority that is far from its own king has
>very good winning chances if his king is safe from perpetual. In this case, the
>classification that I made is useless because the bad can turn into the good one
>because it's far from the king. Take second example: with queens on board white
>can play c3-b3-c4 and create a passer that will be suported by the queen.
>
>4) Minor Pieces endings: These are the endings where the advantage is usually
>enough, but with some exceptions: oposite bishops (as always), Knight vs Good
>Bishop (the side with the bad structure holding the good bishop). If the side
>with the good majority also has bishop pairs he will be winning as the bishop
>pair will suport their advance.
>
>When I have a good pawn majority I usually try to avoid the 1st 3 types of
>endings because it's more difficult to use the advantage (unless I see I'm going
>to get a really winning position). So, in the case of the pawn structures that I
>gave, I use the only open file to trade down heavy pieces.
>
>I also don't go for the endgame if it's unclear that my majority is the best
>one. I make the position clear before I go for it.
>
>I have 2 interesting games I played against IM's on ICC that ilustrate this
>topic:
>
>[Event "ICC 3 0"]
>[Site "Internet Chess Club"]
>[Date "1999.08.22"]
>[Round "-"]
>[White "Gruen"]
>[Black "Helcio"]
>[Result "0-1"]
>[WhiteElo "2477"]
>[BlackElo "2311"]
>[Opening "Scandinavian: Pytel-Wade variation"]
>[ECO "B01"]
>[NIC "SD.02"]
>[Time "07:31:01"]
>[TimeControl "180+0"]
>
>1. e4 d5 2. exd5 Qxd5 3. Nc3 Qd6 4. d4 Nf6 5. Nf3 a6 6. Bc4 b5 7. Bb3 Bb7 8.
>O-O e6 9. Bg5 Nbd7 10. Re1 Be7 11. Qd3 O-O 12. Rad1 Rfd8 13. a3 c5 14. dxc5
>Qxd3 15. Rxd3 Nxc5 16. Rxd8+ Rxd8 17. Ba2 h6 18. Be3 Nd5 19. Nxd5 Bxd5 20.
>Bxd5 Rxd5 21. Kf1 f6 22. Ke2 Kf7 23. Rd1 Rxd1 24. Kxd1 e5 25. Ke2 Ke6 26.
>Nd2 f5 27. f3 Kd5 28. g3 g5 29. b4 Ne6 30. Kd3 Bf6 31. Nb3 Bd8 32. Nc5 Nxc5+
>33. Bxc5 h5 34. h3 g4 35. fxg4 fxg4 36. hxg4 hxg4 37. Be3 e4+ 38. Kc3 Bf6+
>39. Kb3 Be5 40. Bf2 e3 41. Bxe3 Bxg3 42. a4 Be5 43. axb5 axb5 44. c3 g3 45.
>Bg1 Bf4 46. Bb6 g2 47. Bg1 Bg3 48. Kb2 Kc4 49. Kc2 Be1 50. Be3 Bxc3 51. Bc5
>Bxb4 52. Bg1 Bc5 53. Bh2 g1=Q 54. Bxg1 Bxg1 55. Kb2 b4 56. Kc2 b3+ 57. Kb2
>Bd4+ 58. Kb1 Bc3 59. Kc1 Kb4 60. Kb1 Ka3 61. Kc1 Ka2 62. Kd1 b2 63. Kc2 Bf6
>64. Kd3 b1=Q+ 65. Ke3 Qf5 66. Kd2 Kb3 67. Ke2 Qg4+ 68. Ke3 Qd4+ 69. Kf3 Kc3
>70. Ke2 Qd3+ 71. Kf2 Bd4+ 72. Kg2 Qe3 73. Kf1 Qf2# {White checkmated} 0-1
>
>In this first game black only idea in the opening was to get a favorable endgame
>with the better majority(4 vs 3). I used my better control of d file to trade
>heavy pieces, then exchanges some others until I got not only the better
>majority but also the good bishop. After 16. ... Rxd8 a good evaluation must
>say: Black is much better +/- (not +/=). But only after 32. ... Nxc5 I would
>evaluate the position as clearly winning for black.
>
>[Event "ICC 4 0 u"]
>[Site "Internet Chess Club"]
>[Date "1999.06.08"]
>[Round "-"]
>[White "Helcio"]
>[Black "wally"]
>[Result "0-1"]
>[WhiteElo "2405"]
>[BlackElo "2639"]
>[Opening "QGD: Charousek (Petrosian) variation"]
>[ECO "D31"]
>[NIC "QO.14"]
>[Time "09:57:19"]
>[TimeControl "240+0"]
>
>1. d4 d5 2. c4 e6 3. Nc3 Be7 4. Nf3 Nf6 5. Bg5 h6 6. Bxf6 Bxf6 7. e3 O-O 8.
>Qc2 Na6 9. Rd1 c6 10. a3 Qe7 11. h3 Nc7 12. Be2 dxc4 13. Bxc4 e5 14. dxe5
>Bxe5 15. Nxe5 Qxe5 16. O-O Be6 17. Bxe6 Nxe6 18. b4 a5 19. Rd3 axb4 20. axb4
>Rfd8 21. Rxd8+ Rxd8 22. Rd1 Rxd1+ 23. Nxd1 Qb5 24. Qd2 c5 25. bxc5 Qxc5 26.
>e4 b5 27. Ne3 b4 28. g3 Nd4 29. Kg2 Qc3 30. Qa2 b3 31. Qa8+ Kh7 32. Qb7 b2
>33. h4 Qb3 34. Qd7 b1=Q 35. Qxd4 Q3d3 {White resigns} 0-1
>
>This one I lost because I evaluated wrongly the position after 13. ... e5
>I thought I would get the better majority but this is completely wrong because
>It's hard to exchange queens and although I had a3,b4 against a7,b7,c6 The
>position was worse for me. Better would be 14. o-o and although after 14. ...
>exd4 15 exd4 white has an isolated pawn he is better because he has more active
>pieces and can play against c8 bishop(a bad bishop). It's clear that after 22.
>Rxd1+ Black is winning easily... :(
this is always a risk... particularly if you are thinking majority at move
13. :) Because the middlegame is between you and victory, as always. I'm
going to hold 'candidate passer' scores way down until material begins to thin
out significantly, although I will always evaluate passed pawns highly. But
even my normal passed pawn scoring is dependent on material present on the
board..
>
>When you understand well a theme like this one, you improve a lot your game and
>so will if you implement it on a chess program. Perhaps an aproach to this would
>be to have an aditional flag to say:endgame is good, exchange this type of
>pieces... If you should use it in the evaluation and if so, what should be it
>height on it is a more complex question.
>
>Helcio Alexandre
Tough questions, to be sure... "outside passed pawn" was not really hard after
I caught the obvious exceptions (opponent has a protected passer, etc.) But
adding this is harder becuase this has to work in harmony with all the other
positional endgame terms I have in Crafty, and if they get out of whack, it
starts playing like a bozo, quickly.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.