We were held to a draw (+1=2-1) by a much-outgraded Hammerheads team in the third round of matches at Battersea Labour Club.
The result means we have 7/9 match points (3pts for a won match, 1pt for a drawn one) and 9.5/12 game points. But each division is being run as a nine-round all-play-all, so there is a long way to go.
My game on board one got down to a minor-piece ending that either I seriously misjudged or analysis engines do not understand.
 |
| White to make his 27th move in Spanton (171) - Claudio Moni (-) |
*****
*****
*****
*****
White has the superior pawn-structure - it will be relatively easy to make a passed pawn on the queenside, but very difficult for Black to do likewise on the kingside. White also has the better minor piece for coping with rival pawn-majorities.
Black has more kingside space and the more-centralised king. He also has practical chances based on the knight's forking abilities and the fact that White's bishop is restricted to dark squares.
I was fairly confident this should be a win for White, so I was very surprised on quickly putting the game into ChessBase when I got home in the early hours - there was a foul-up on the Waterloo & City line - to find Stockfish10 and Komodo9 have the initial position as equal.
As I inputted the moves, there were points where the engines reckoned Black was better, so it will be interesting, at least for me, to try to discover the truth today, now I have more time to devote to it.
27.f3
Having typed the last few paragraphs while I left the engines running, I returned to the diagram position to find they have changed their view - Stockfish10 has White about 0.4 pawns ahead; Komodo9 about 0.25.
However, neither engine likes my move. Stockfish10 recommends 27.f4!? This is very controversial. True, it gains space on the kingside, but putting pawns on the same-coloured squares as the bishop is very much what White wants to avoid in this sort of ending. White's ideal set-up, I believe, is to have the pawns covering light squares and the bishop covering the gaps, ie the dark squares, to prevent the knight attacking the pawns.
Komodo9 wants to start centralising the king with 27.Kf1 while retaining maximum flexibility with the pawns.
27...Nc6 28.Bc3 b5
The engines do not like this gain of space. Instead they suggest the committal 28...f4!? or the natural 28...Ke6.
29.Kf2
Stockfish10 gives White a slight edge after 29.g3; Komodo9 after 29.f4!? Again I reckon it is a balancing act between the desirability of gaining space and the drawback of putting pawns on dark squares.
29...Ke6
Both engines greeted my last move by flashing yellow (meaning "dubious") and then red (meaning "bad"), settling on the latter. They give 29...f4, at first reckoning Black has a slight edge. But this evaluation soon changes to equal. I then backtracked the last few moves, before going through them again, and this time the engines see nothing wrong with 29.Kf2, keeping to a steady green.
30.Ke3 g5 31.Bd4
Once again the engines strongly prefer the move f4!?
31...a6 32.a3
A dark square, but I felt this was the best way to deal with the threat of ...Nb4.
32...Ne7 33.Kd3
White's winning plan is simple enough - create a passer on the queenside while ensuring Black cannot do anything dramatic on the kingside.
33...Nd5 34.Be3 h5
Most humans, I am sure, would reject 34...Nxe3?? out of hand, and rightly so. However, Komodo9 at first rates the position as equal, before changing to plus-over-minus and eventually to winning for White. Stockfish10 is also slower than I would expect to realise the pawn-ending is hopeless for Black, who will find it impossible to create a passed pawn.
35.b3
CM offered a draw while thinking about his next move.
35...Ke5 36.c3?!
Overly cautious. I could have played c4 straight away, but maybe nothing is spoilt by the text.
36...g4
The engines prefer 36...h4, but that does not really change anything.
More promising, or at least trickier, is 36...Ne7!?, intending to attack White's g pawn, eg 37.c4 bxc4+ 38.Kxc4 Ng6.
 |
| Analysis diagram - position after 38...Ng6 |
*****
*****
*****
*****
We can dismiss 39.g3?! as 39...f4 lets Black undouble his pawns, and the position seems to be drawn.
For a long time the engines want White to play 39.b4!?, and they keep coming back to this move, but after 39...Nh4 a win is far from clear. One line given by Stockfish10, with both engines' evaluations changing frequently, goes 40.Kc5 Nxg2 41.Bf2 Kf4 42.a4 Kxf3 43.Bg1 Ne3 44.b5 axb5 45.axb5 f4 46.b6 Kg2 47.b7 Kxg1 48.b8=Q Kxh2, when we have the rarely (extremely rarely) seen ending of queen versus knight and four pawns - a dead draw, according to the engines.
That would be somewhat difficult, to say the least, to work out over the board, so perhaps the most sensible continuation is 39.Bf2, which seems to give winning chances after 39...Nf4 40.g3 Nh3 41.Bd4+ (Komodo9 has White almost a full pawn ahead; Stockfish10 reckons White's advantage is worth less than a tenth of a pawn). Black has other options, but in each case the question is whether White is winning or Black can somehow scrape a draw. For practical purposes, which admittedly is not the same as the ultimate truth, White is very likely to win against 36...Ne7!?
37.a4?
A mistake that should have cost me the win. I rejected the natural 37.c4 because after 37...Nxe3? 38.Kxe3 bxc4 39.bxc4 f4+, I thought White had to play 40.Kd2??, losing the c pawn. Somehow, in my mind's eye, I thought the winning 40.Kd3 was illegal.
37...Kd6 38.axb5 axb5 39.c4 bxc4+ 40.bxc4 f4??
The first move all three of us (Stockfish10, Komodo9 and I) agree is bad. Black has a draw after 40...Ne7, eg 41.Kd4 Ng6, intending ...Nh4. Komodo9 gives White either a slight edge or the upper hand after 41.f4, but again 41...Ng6 is a complete answer, eg 42.g3 (42.Kd4 Nh4 43.g3 Nf3+) h4 43.Bd4 h3, when White cannot even win one of Black's doubled pawns (44.Bxf6? Nxf4+!), not that winning the f6 pawn would win for White in any case.
41.Bd2
41.cxd5 also wins.
41...Nb6 42.Bxf4+ Kc6
42...Kc5? 43.Be3+.
43.fxg4 hxg4 44.Kd4 Nd7 45.h4 1-0 (50 moves)
So what are my conclusions?
1. This ending really is one engines find very difficult, presumably because of the myriad of non-forcing possibilities.
2. I was probably right to regard White as having a close-to-winning position in the first diagram.
Major caveat: my understanding of how White should play on the kingside, ie holding back pawns to avoid fixing them on dark squares, may be wrong - the engines' insistence on gaining space with f4 may be the lesser evil.
My updated Battersea statistics for 2018-19
Event...Colour...Grade...Opponent's Grade...Result
CLL........B..........167...….........196...............…D
LL...…….B...…...167...………..159...………….D
CLL...…..B...…...167...………..161...………….W
CLL...…..B...…...167...………..190...………….D
LL...…….W...…..167...………..161...….………W
CLL...…..W...…..167...………..148...………….D
CLL...…..W...…..167...………..165...………….W
CLL...…..W...…..167...………..160...………….D
CLL...…..W...…..167...………..159...………….D
LL...…….B...…...167...………..168...………….D
LL...…….W...…..171...………..159...………….W
CLL...…..B...…...171...………..198...……….….L
CLL...…..B...…...171...………..169...…………..L
CLL...…..B...…...171...………..196...…………..L
LL...…….B...…...171...………..182...…………..D
CLL...…..W...…..171...………..189...…………..L
LL...…….B...…...171...………..178...…………..L
CLL...…..B...…...171...………..164...…………..D
LL...…….B...…...171...………..188...………….W
CLL...…..B...…...171...………..200...…………..L
LL...…….B...…...171...………..169...…………..L
CLL...…..W...…..171...………..186...…………..D
CLL...…..B...…...171...………..153...…………..D
LL...……W...…...171...………..188...…………..L
LL...……W...…...171...………..159...…………..L
LL...……W...…...171...………..153...…………..L
CLL...…..B...…...171...…….….172...……….....D
LL............B.........171................139..................W
LL...….….B...…..171...…….….214...……....….L
LL...……..B...…..171...………..173...……….….L
LL...…….W...…..171...………..166...…………..L
SL...……..B...…..171...………..167...………….W
LL...…….W...…..171...………..122...………….W
SL...……..B...…..171...………..159...………….W
SL...…….W...…..171...………..140(est)………W
Overall this season for Battersea I have scored +10=12-13 for a grading performance of 166.
In season 2017-18 I scored +10=8-9 for a grading performance of 175.
CLL - Central London League; LL - London League; SL - Summer League
No comments:
Post a Comment