Friday, January 4, 2008

Attn: Ending Experts! Is this a draw?

In my 9+10+9+12=40 post I made reference to my third round loss, but wanted to deal with the position in a separate post. The crucial position is rich with possibilities, and I didn't want it to get buried amongst my long winded journey through the mind fields of pint sized chess monsters. Here is the game in it's entirety.


The way I played it out I thought I was dealing with a rook pawn ending. Unfortunately with pawns on the other side I don't have stalemate possibilities. On move 58 we reach a typical rook pawn stalemate position, except I have pawn moves on the king side.

At move 50 for black we reached the position below. My oppenent felt I could have drawn if I had played 50...g5. I had done a count and knew he was promoting before me, but pawn is advanced enought that it doesn't seem like he can stop me from promoting even though he promotes 2 moves earlier then I possibly could.

Here is some of the analysis I looked at with the aid of Fritz. After this little exercise I'm convinced Fritz knows jack shit about endings. From the moment I examined 50...g5, Fritz rated the position +- 7.41. At that point I didn't feel so bad about playing the line I did. A loss is a loss. But look what happens.

50... g5 51. hxg5+ hxg5 52. fxg5+ Kxg5 53. Kb5 Kf4 54. Kxa5 Kf3 55. Kb5 Kxf2 56. a5 f4 57. a6 f3 58. a7 Ke2 59. a8=Q f2 60. Qe4+ Kd1 61. Qf3+ Ke1 62. Qe3+ Kf1 63. Kc4 Kg2 64. Qe4+Kh2 65. Qh4+ Kg2 66. Qg4+ Kh1 67. Qf5 After this move Fritz gives White+- 4.62 But look at the evaluation one move later after 67...Kg2 = 0.25 .

I looked at a few different variations, and all of them would give White a huge advantage at first and as long as Black guarded the pawn without going Kf1 Fritz would then rate the position equal.

Here's the million dollar question: Can black hold after 50...g5?

Gentleman, start your engines.


Ryan Emmett said...

It seems to me that there is no way for Black to save the game.

After 1...g5 2.hxg5 hxg5 then NOT 3.fxg5 giving Black's King easier access to the remaining White pawns, but instead 3.Kb5 and Black is lost e.g. 3...gxf4 4.Kxa5 Ke5 5.Kb5 Ke4 6.a5 Kf3 7.a6 Kxf2 8.a7 Ke2 9.a8Q

My Fritz 11 'thinks' the game is drawn (evaluation 0.00) after 3.fxg5.

BTW - I thought Fritz 7 came with endgame databases? In the line you gave my Fritz consults it's 'tablebases' after 55...Kxf2 and declares it a definite draw.

Polly said...

This is Fritz 5.32 that comes with Chessbase 9. I haven't tried just putting the position into Fritz 10 and seeing what it comes up with.

Fritz liked your line better, and that clearly is an easier win for white.

OK now I don't feel so bad.

Fritz also doesn't take into account the clock issues. At move 50 we both were under a minute.

liquideggproduct said...

Try 54. ... Kc6 instead of Kb8. I think you can get the draw that way.

liquideggproduct said...

Clarification: in the original game, move 54. The idea is that you trap his King on the side, and only after he uses the f2-f3 tempo, slide into b8.

Temposchlucker said...

Rybka plus Nalimov tablebase give mate in 29 from the diagram.
The main line runs as follows:

1. ... Ke6 2.Kb5 Kd5 3.Kxa5 Kc5 4.Ka6 Kc6 5.a5 h5 6.Ka7 Kc7 7.a6 Kc6 8.Kb8 Kb6 9.a7 after which mate in 21 follows.

After 1. ... g5 it is mate in 24, which is 5 moves faster.

The main line goes as follows:

g5 2.hxg5+ hxg5 3.Kb5 gxf4 4.Kxa5 Ke5 5.Kb5 Kd6 6.Kb6 Kd7 7.Kb7 Kd6 8.a5 and mate in 17

In the variation you give, white gives the game away with immediate draws after 52.fxg5
In stead Kb5 is necessary. As Ryan already pointed out.

Soapstone said...

My sense is that your endgame is lost after 32...Rf5. The pawn with the rook wind at its back running to c7 ensures that your king will be poorly placed compared to your opponent's and that's the difference in the ending.

In the original game, 50...Ke6, chasing the king to the queenside is futile. After capturing your a-pawn, White creates the "Advantage of the outside passer" which isn't so dangerous in itself, but it either queens or distracts you long enough for White to go back to the kingside and eat all the remaining pawns. Your opponent didn't quite do this right, but luckily he had zugzwang on the kingside pawns that caused you to have to allow a passer.

Instead of chasing, 50...g5 is a decent chance for a swindle. I agree with ryan that 52.fxg5+ in the Fritz analysis line is definitely drawn. The ending position is a book Queen versus bishop pawn draw. Instead of 64.Qe4+, White should play 64.Qg5+ Kh2 65.Qf4+ Kg2 66.Qg4+ Kh2 67.Qf3 Kg1 68.Qg3+ and Black must play Kh1! because Qxf2 stalemates. White doesn't have time to bring his king closer while the f2 pawn threatens to promote.

Did the game really end on 59...hxg5? I know you were already frazzled at the end of time control, but you could have played on a few more moves hoping for 60.hxg5 f4 61.g6 f3 62.g7?? stalemate! If he plays 60.h5, or releases his a-pawn with Kb5 for your king to capture then there's virtually no hope left.

BlunderProne said...

If I have trouble evaluating an end game and trusting Fritzy... I like the following website:

I already tried pluggin in the FEN of your position and it doesn't appear to be in the data base. Future reference this might be a useful link for you?

Polly said...

LEP: I'm not sure your idea works since as long as he doesn't push the pawn all the way down to a7, I will eventually be forced to give ground or play g5. I think it simply transposes to Ryan's line.

After 54. Kc6, 55. Ka7 Kc7, 56. a6 Kc8 57. Kb6 Kb8 58. a7+ Ka8 59. Ka6

Polly said...

Soap: Actually I resigned after 59. fxg5. When I reconstructed the last few moves I accidently put hxg5 in the Mon Roi. Since it was the last move I put in I can't "retract" it. I never actually played the move, but I realized if I do play it he pushes h5 and it's mate in 4.

Blunder: Thanks for the website addy. I'll keep that in mind the next time I have an ending taht confounds me.

likesforests said...

I don't think Rybka or Fritz is necessary for this position. :)

1.g5 2.h5!?

Now there's only one line to calculate and every move of the race is forced. How's it end?

White needs 7 moves to promote... K->b5,a5,b6 P->a5,a6,a7,a8.

Black needs 9 moves to promote... P->f4 K->g5,h5,g4 P->h5,h4,h3,h2,h1.

So a8=Q h2 Qg2+ and White has a very simple mop-up operation!

Temposchlucker said...

LF, I don't think Rybka or Fritz is necessary for this position.
From Rybka, Fritz and Likesforests the first two are easier accessable:)

If black plays g3 in stead of h2 then to me it is not immediate evident that the mop up is going to be all that easy.

likesforests said...

tempo, g3 is not possible after the sequence I gave. Maybe my shorthand was too cryptic? 1...g5 2.h5 gxf4 3.Kb5 Kg5 4.Kxa5 Kxh5 5.Kb6 Kg4 6.a5 h5 7.a6 h4 8.a7 h3 9.a8=Q.

If you meant 9...f3!? then 10.Qg8+ followed by 11.Qg3 and Black can resign.

liquideggproduct said...

I'm such a putz. For some reason, I was imagining you couldn't play g5 anymore.

Polly said...

LEP: Welcome to my world of imagination chess. I see things that aren't there, and don't see things taht really are there.

takchess said...

I must say that is a nice crew coming to help you analyse. Interesting stuff, and I think we can crown Like Forest KING OF THE ENDGAMES.