Saturday, 21 December 2024

What Do The Engines Say? 2. French Defence

MODERN computer programs are revolutionising how we approach openings.
This first became noticeable when supposedly refuted moves started making a comeback as engines showed resources that had evaded the world's top players.
But more recently, as engines grow in strength, there has been a trend in the opposite direction, at least at the highest level.
Grandmaster Nigel Davies summed this up in a post at chesspublishing.com, where he recounted a conversation with Canadian GM Kevin Spraggett, who "opined to me that ancient openings are making a comeback because engines are finding flaws in the more risky ones."
Davies added: "I think there is a lot of truth in this, which is confirmed by the migration of top players to strictly classical openings."
Engines are still far from the stage of knowing all there is too known about openings.
Their play in that stage of the game is not on a par with their abilities in tactical middlegames.
We cannot expect engines to tell us move-by-move how to play our favourite opening variations,
But their opinions on what we should be looking at - the moves that show the most promise - are certainly well worth considering.
So in this series I will be looking at what the engines say about mainline opening positions.

WHEN hot-favourite Gukesh Dommaraju sat down with the white pieces for the first game of the world championship match, he was probably not expecting the champion, Liren Ding, to play the French Defence.
Gukesh met 1.e4 e6 2.d4 d5 with 3.Nc3, and, when Ding replied 3...Nf6, continued with the Steinitz Variation, ie 4.e5.
Stockfish17 and Dragon1 fluctuate between 3.Nc3 and 3.e5 (the fashionable - after a long time in the doldrums - Advance Variation), but come to settle on the former, albeit marginally.
After 3...Nf6 Stockfish17 agrees with 4.e5, but Dragon1, although at first also preferring that move, comes to give the edge to the more popular 4.Bg5.
Both engines reckon their move gives White a slight edge, while claiming the other move leaves the game equal.
However, allow Dragon1 a lot more time to consider the position, and it too comes to prefer 4.e5, albeit marginally.

Friday, 20 December 2024

What Do The Engines Say? 1. Evans' Gambit

MODERN computer programs are revolutionising how we approach openings.
This first became noticeable when supposedly-refuted moves started making a comeback as engines showed resources that had evaded the world's top players.
But more recently, as engines grow in strength, there has been a trend in the opposite direction, at least at the highest level.
Grandmaster Nigel Davies summed this up in a post at chesspublishing.com, where he recounted a conversation with Canadian GM Kevin Spraggett, who "opined to me that ancient openings are making a comeback because engines are finding flaws in the more risky ones."
Davies added: "I think there is a lot of truth in this, which is confirmed by the migration of top players to strictly classical openings."
Engines are still far from the stage of knowing all there is to know about openings.
Their play in that stage of the game is not on a par with their abilities in tactical middlegames.
So we cannot expect engines to tell us move-by-move how to play our favourite opening variations,
But their advice on what we should be looking at - the moves that show the most promise - is certainly well worth considering.
So in this series I will be looking at what the engines say about mainline opening positions, starting with the Evans Gambit.

After 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.b4 Stockfish17 and Dragon1 agree the pawn should be taken (with the bishop).
Position after Black takes the pawn
*****
*****
*****
*****
*****
Dragon1 reckons White has fair compensation, although it believes White no longer has the advantage that comes with moving first, while Stockfish17 already gives Black a slight edge.
However, the situation is not clear, as is shown by the engines' reaction to White's follow-up, 5.c3.
Stockfish17 recommends looking at 5...Ba5, which is the main continuation in ChessBase's 2025 Mega database, but Dragon1 fluctuates between that and the second-most popular move, 5...Be7.
One thing the engines are agreed on, however, is that 5...Bc5, which used to be touted in books as roughly as good as 5...Ba5 (5...Be7 rarely got a look-in), helps White, with Stockfish17 claiming it leaves White slightly better (Dragon1 is not so sure).

Thursday, 19 December 2024

Fiery Dragon

HERE is an exciting and instructive game from Mikhail Botvinnik's One Hundred Selected Games.
Notes in italics are from the book.

Ilya Kan - Mikhail Botvinnik
Moscow 1936
Sicilian Dragon
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 g6 6.Be2 Bg7 7.Be3 Nc6 8.Nb3!?
Am idea deserving serious attention. White has in mind an immediate pawn-attack on the kingside, while Black has no better continuation than 8...Be6 and 9...0-0, as otherwise White will castle short, and then a well-known variation unfavourable to Black will result.
More popular than the text were, and still are, 8.Qd2 and especially 8.0-0. For what it is worth, Stockfish17 very marginally prefers the former, while Dragon1 favours the text.
8...Be6
It is hard to know what Botvinnik's "well-known variation unfavourable to Black" may be. Certainly nothing obvious jumps out from ChessBase's 2024 Mega database.
9.f4 0-0 10.g4!?
How should Black meet White's pawn-storm?
*****
*****
*****
*****
*****
10...Na5?!
In reply to White's flank attack, Black should have undertaken an active counterattack in the centre by 10...d5! Now White's attack grows dangerous.
The text is by far the commonest continuation in Mega24, but is strongly disliked by Stockfish17 and Dragon1. They suggest 10...Rc8 or 10...a5, but reckon Botvinnik's recommendation is a mistake thanks to the reply 11.f5. This was played later the same year at Nottingham in the game Alexander Alekhine - Botvinnik, which continued 11...Bc8!? 12.exd5 Nb4, after which the engines reckon 13.Bf3 gives White the upper hand, as subsequent games seemed to show.
11.g5 Ne8
This is unquestionably better than 11...Nd7 as afterwards the knight can occupy g7, where it is useful for defence.
The latter move is almost five times more popular in Mega24, and is the top choice of Stockfish17, whereas Dragon1 agrees with Botvinnik's assessment.
12.Qd2
White determines the queen's position prematurely. The correct move is 12.Bd4 at once, as [Jan] Foltys played against [Erich] Eliskases (PodÄ›brady, Czechoslovakia, 1936).
The engines marginally prefer Kan's move.
12...Rc8!?
Sacrificing the a7 pawn. White refuses the sacrifice, for after 13...Bc4 it would lose several important tempi and relinquish the initiative to Black.
The engines agree 13.Bxa7? is a mistake, but reckon the correct reply is 13...Bxc3!? 14.bxc3 Nc4 or the immediate 13..Nc4. The problem with 13...Bc4, according to them, is the simple 14.Bd4, when they reckon White is at least equal.
13.Bd4 Nc4
Stronger is 13...Bc4, eg 14.0-0-0 is followed by 14...Bxe2 15.Nxe2 (or 15.Qxe2) Nc4, and the knight's strong position gives Black opportunities of counterattack.
The engines prefer the text, but agree with Botvinnik's analysis of his main variation, with the major exception that they reckon meeting 15.Qxe2 with ...Nc4 gives White a winning attack starting with 16.h4. Instead they recommend 15...e5!?, claiming complete equality.
14.Bxc4 Rxc4 15.0-0-0 Qd7 16.Qd3 Rc8!?
After this passive move the initiative passes to White. Black should have continued 16...b5 with an approximately equal game.
The engines agree 16...b5 is a little better, but reckon White anyway has an edge.
17.h4 Bg4
Black must prevent the opening of the h file.
18.Rd2
Retreat to e1 would be more exact. The rook's position on the same diagonal as the king provides Black with combinative possibilities.
18...b6?!
Botvinnik does not comment on this defensive move, but it seems a serious mistake. The engines give 18...f5, albeit awarding White a slight edge.
19.Nd5
If 19.Bxg7 then 19...Nxg7 20.Nd5 Be6 21.Nd4 Rc5, and Black has fully equal play.
The engines prefer 19.Bxg7, their point being to meet 19...Nxg7 with 20.Rg1!?, eg 20...f5 21.e5 Rfd8 22.Qd5+ Qe6 23.Nd4!? Qxd5 24.Nxd5, claiming White has the upper hand.
19...e5
With passive play Black would be slowly asphyxiated after f5.
20.Bc3
Apparently 20.f5 is also without danger to Black, because of 20...exd4 21.f6 Nxf6 22.gxf6 Bh6 23.Ne7+ Kh8 24.Nxc8 Rxc8.
The engines agree with the assessment of the result of that line, but reckon sacrificing the exchange with 23.Kb1!? Bxd2 24.Qxd2 gives White a slight edge.
20...f5?!
The engines reckon Black needed to play 20...Nc7 or possibly 20...f6.
Position after 20...f5?!
*****
*****
*****
*****
*****
*****
An extremely sharp and complex position has arisen. To 21.fxe5 Black can simply reply 21...dxe5.
The engines continue 22.Qa6! with what seems a winning advantage as Black has apparently unsolvable problems saving his queen without losing material, eg 22...Nd6 23.Bb4 Rcd8 24.Qa3! Nc4 25.Ne7+ Qxe7 (or 25...Kf7 26.Rxd7 Nxa3 27.Rb7! etc) 26.Bxe7 Nxa3 27.Rxd8, winning the exchange.
Possibly strongest of all now for White is 21.exf5 Bxf5 22.Qe2, and White preserves his knight at d5. The continuation White chooses gives the advantage to Black.
21.gxf6?! Nxf6 22.Nxf6+ Rxf6 23.Qd5+?
Nothing in 23.fxe5 Rf3 24.Qd5+ Be6 25.Qxd6 Bh6 26.Nd4 Rh3!, and White does not get adequate compensation for the exchange.
The engines reckon this line gives White full compensation.
23...Rf7!
The move that decides the game! If now 24.Qxd6 then 24...Qxd6 25.Rxd6 exf4, or 24.fxe5 Be6 25.Qxd6 Bh6, and White is in a bad way.
Now White makes a pawn sacrifice, which gives him nothing and only speeds up his loss.
24.h5 gxh5 25.Rg1
In sacrificing the pawn White evidently overlooked that after 25.fxe5 dxe5 26.Qxd7 Bxd7 27.Rxh5 Rf1+ 28.Rd1 Rxd1 29.Kxd1 Bg4+ he is a rook down.
25...Qc6 26.fxe5 dxe5!?
The strong passed h pawn secures Black the win. So there is no sense in complicating the game by winning the exchange at the price of laying bare his king's position.
In other words, Botvinnik is saying that 26...Bh6, while it may be objectively stronger, as the engines agree, increases the chances of Black going wrong.
27.Qxc6 Rxc6 28.Rd5 Rcf6 29.Nd2
The continuation 29.Bxe5 leads to loss of material because of 29...Rf1+ 30.Rxf1 Rxf1+ 31.Kd2 Rd1+.
29...Bh6 30.b3 Rf2 31.Kb2
31.Rxe5 does not save the game because of 31...Rh2 ( to defend the h5 pawn), followed by 32...Rff2.
31...Re2?
Botvinnik does not comment on this. It seems both players missed that this throws away Black's advantage, which would remain more-or-less decisive, according to the engines, after 31...Bg5 or 31...b5.
32.Nc4 Rff2 33.Rd8+ Bf8
The engines slightly prefer moving the king, or 33...Rf8.
How should White continue?
*****
*****
*****
*****
*****
34.b4?
White's game is lost, but even so he should have chosen 34.Rc8, although here also the advance of the h pawn is decisive.
The engines reckon 34.Rc8?? is even worse than the text, but that 34.Kb1 gives complete equality, and 34.Be1! may even lead a to a tiny edge, eg 34...Rxc2+ 35.Kb1 Rxc4! (moving the f2 rook along White's second rank allows 36.Rf1) 36.bxc4.
34...Rxc2+ 35.Kb3 Rg2 36.Rf1?
Better are 36.Rxg2 and 36.Bxe5!
36...Rcf2 37.Rxf2 Rxf2 38.Nxe5 Be6+ 39.Ka4 Rxa2+ 40.Kb5 Rc2 41.Rd3 h4 42.Bd4 Bg7 43.Ra3 h3 44.Rxa7 h2 45.Ra8+ Bc8 46.Ra1 Bh3 47.Nf3 Bxd4 48.Nxd4 Rf2 0-1

Wednesday, 18 December 2024

Summing Up Coulsdon

MY score of +2=0-2 (I also took a half-point bye) lost 2.2 ECF elo and 19.8 Fide elo.

Tuesday, 17 December 2024

Coulsdon Round Five

DOUBLE-upfloated against a junior (born 2005).

Henry Adams (2119 ECF/1991 Fide) - Spanton (1965 ECF/2012 Fide)
Catalan
1.d4 d5 2.Nf3 e6 3.g3 c5 4.Bg2 Nc6 5.0-0 Nf6 6.c4 dxc4
This position occurs 7,246 times in ChessBase's 2025 Mega database
*****
*****
*****
*****
*****
7.Qa4
This is the commonest continuation in Mega25, but Stockfish17 and Dragon1 prefer 7.Ne5!? or 7.dxc5!?
7...Bd7 8.dxc5!? Bxc5 9.Qxc4
How should Black proceed?
*****
*****
*****
*****
*****
9...Qe7!?
The engines prefer 9...Be7 or 9...Bb6, which brings to mind advice from chess writer and first world correspondence champion, Cecil Purdy: if a piece is attacked, don't defend it - move it.
10.Nc3
The engines reckon 10.Bg5!? gives a slight edge.
10...Rc8?!
Black maintains equality with 10...h6!?, according to the engines.
11.Bg5 Bb6?
Much better, according to the engines, are 11...0-0 and 11...h6!?, even though both moves can be met by 12.Bxf6, when 12...gxf6 is forced.
12.Qh4
Black is losing
*****
*****
*****
*****
*****
12...Qc5?!
Best may be 12...Bd8!?
13.Bxf6 gxf6 14.Ne4!
This is even stronger than 14.Qxf6.
14...Qe7 15.Nxf6+ Kf8??
Better, but still horrible, is 15...Kd8 16.Rfd1 Rc7.
16.Qh6#

Monday, 16 December 2024

Coulsdon Round Four

DOWNFLOATED against a junior (born 2009).

Spanton (1965 ECF/2012 Fide) - Albert Eldrige (1851 ECF)
Blackmar-Diemer Gambit/Caro-Kann
1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.Bc4!?
4.Bc4!? scores 52% in ChessBase's 2025 Mega database, the same as 4.Nxe4 but two percentage points higher than 4.f3!?
*****
*****
*****
*****
*****
Then again 4.Nge2 scores 67%, which presumably is largely down to its surprise value.
4...Nf6 5.f3!?
How should Black respond?
*****
*****
*****
*****
*****
5...c5!?
Probably a novelty. The mainline in Mega25 runs 5...exf3 6.Nxf3 Bf5 (6...e6!? is also very popular) 7.0-0 e6, with at least a slight edge for Black, according to Stockfish17 and Dragon1.
How should White proceed?
*****
*****
*****
*****
*****
6.d5!?
The engines agree this is best, although it, at least temporarily, blocks the light-square bishop. Stockfish17's second choice is 6.dxc5!?, continuing 6...Qxd1+ 7.Nxd1!? e6!?, awarding Black a slight edge. Dragon1's second choice, for a while, before switching to Stockfish17's preference, is 6.Bf4!?, when 6...cxd4 7.Nb5 Na6 8.fxe4!? Qa5+!? 9.Bd2 Qb6 10.e5 Bd7!? 11.a4 gives a sharp position that may be dynamically equal, but there are many plausible alternatives along the way.
6...Bf5?!
The bishop is vulnerable here. Stockfish17 suggests 6...Nbd7 or 6...a6!?, while Dragon1 advises acceptance with 6...exf3.
7.g4!? Bd7!?
After 7...Bg6 8.g5 Nfd7 9.fxe4 White is also doing well.
8.g5 Ng8?!
Probably better is 8...Nh5, meeting 9.fxe4 with 9...g6.
9.fxe4 a6 10.a4 h6?!
The engines suggest trying to block the centre with 10...e5, their point being 11.dxe6 Bxe6 12.Qxd8+ Kxd8 13.Bxe6 fxe6 leaves White on top - positionally winning, according to the engines - but with the black king in little danger of being mated. However they reckon White does better getting on with development, eg 11.Qf3.
11.Qf3 e6?
With the white queen off the d file, moving the e pawn is simply a mistake. The engines suggest the consistent 11...hxg5.
12.dxe6 Bxe6 13.Bxe6 fxe6 14.Qh5+ Kd7 15.Bf4 Kc8 16.Rd1 Qe7
Hardly much of an improvement, if at all, is 16...Nd7 17.Nf3 with Ne5 to come.
17.Bd6 Qxd6 18.Rxd6 Bxd6 19.Qe8+ Kc7 20.Qxe6
White has won queen and pawn for rook and bishop, giving a huge advantage, according to the engines (about +8.00), but with practical difficulties arising from the white king's lack of a safe haven 
*****
*****
*****
*****
*****
20...Nd7 21.Nd5+ Kc6 22.g6?!
Much better, according to the engines, is going after the black king with 22.b4 or developing with 22.Nf3.
22...Ngf6 23.Nxf6 Nxf6 24.e5 Rad8 25.Nf3 Rhe8 26.Qf5!?
The engines prefer 26.Qc4 or 26.Qb3.
26...Nd7 27.e6 Nf8 28.0-0 Rxe6
White's extra pawn has gone, but White is still winning, according to the engines
*****
*****
*****
*****
*****
29.Nh4 Be7 30.Qf3+ Kb6?!
Probably better is 30...Kc7 as the text allows White a simplifying continuation that removes danger from the white king.
31.Qb3+ Ka7 32.Rxf8! Re1+!?
This is best, according to the engines.
33.Rf1
Not 33.Kf2?? Bxh4+.
33...Rxf1+ 34.Kxf1 Bxh4
How would you assess this ending?
*****
*****
*****
*****
*****
White can win a pawn, which I assume means the ending is winning, although in queen-versus-rook-and-bishop endgames there is always the possibility of the weaker side setting up a fortress.
35.Qc3 Bf6 36.Qxc5+ Kb8 37.b3 Rc8 38.Qd6+ Ka8 39.c4 Rd8 40.Qe6 Rf8 41.Kg2 Rd8 42.c5 Kb8
One line illustrative of White's chances runs 42...Rd2+ 43.Kf3 Rxh2 44.Qxf6!? gxf6 45.g7 Rh3+ 46.Kg2 etc.
43.c6 Kc7 44.cxb7 Kxb7 45.a5 Ra8
The only way to defend the a6 pawn.
The game finished:
46.Qb6+ Kc8 47.Kf3 Bd8 48.Qc5+ Kb7 49.Qd5+ Ka7 50.Qd7+ 1-0

Sunday, 15 December 2024

Coulsdon Round Three

AFTER taking a half-point bye in round two, I was upfloated against a junior (born 2011).

Venkata Kilambi (1975 ECF/1879 Fide) - Spanton (1965 ECF/2012 Fide)
Colle
1.Nf3 d5 2.e3 c5 3.d4 e6
This position occurs 3,392 times in ChessBase's 2025 Mega database, a common move-order being 1.d4 d5 2.Nf3 e6 3.e3 c5
*****
*****
*****
*****
*****
4.a3!?
But this continuation occurs just 20 times, although thanks to various move-orders the resulting position appears 68 times in Mega25. The idea will soon become clear.
4...Nc6 5.Be2 Nc6 6.dxc5!?
Stockfish17 and Dragon1 like this, along with 6.0-0. After the latter, the most popular move in Mega25 is 6...Bd6!?, but the engines are not keen on that, claiming White gains a slight edge with 7.dxc5 Bxc5 8.b4. Instead they recommend 6...cxd4 or 6...c4!?
6...Bxc5 7.b4 Bd6 8.Bb2 0-0 9.0-0
Now both sides have castled, how would you assess the position?
*****
*****
*****
*****
*****
Black has a 2-1 majority on the two centre files, but may have difficulty getting the black light-square bishop into play. White's light-square bishop is also passive, but the white dark-square bishop is probably the game's best minor piece. The engines reckon the position is equal.
9...Qc7!?
How should White continue?
*****
*****
*****
*****
*****
10.Nc3?!
VK criticised this afterwards, and it does seem strange to block both White's active bishop and backward c pawn. The known move in the diagrammed position is 10.Nbd2, which looks more natural, but allows Black to reply 10...e5. A critical question is whether White should capture on f6. After 10.Bxf6 gxf6 the black kingside is certainly weakened, but it is not clear White can exploit this in the face of Black's strong centre and bishop-pair. Perhaps White should have attacked the black centre with 10.c4.
10...a6
Preventing Nb5. The engines reckon Black is at least slightly better.
11.h3!?
This may look slow, but it is the engines' top choice along with 11.Bd3!?
11...Rfd8 12.b5 Na5 13.Nd4!?
The engines dislike this, suggesting 13.a4 or 13.Rb1.
13...Nc4 14.Bxc4 dxc4
The engines, particularly Stockfish17, prefer recapturing with the queen.
15.Qe2 e5 16.Nf3 Bd7 17.a4 axb5 18.axb5 Qc5
What should White play?
*****
*****
*****
*****
*****
19.Rab1?
This indirectly defends b5, but misses the other point of Black's last move. The engines suggest 19.Nd2, or exchanging on a8 and then playing Nd2.
19...e4
Nd2 would have prevented this, but now it has come with tempo.
20.Nd2 Qe5 21.f4 exf3 22.Rxf3?
The other two piece-recaptures are better, eg 22.Nxf3 Qg3 23.Qf2 Qxf2+ 24.Rxf2, although White's isolated e pawn and Black's bishop-pair add up to the upper hand for the latter, according to the engines. I was leaning towards meeting 22.Nxf3 with 22...Qh5, which the engines reckon is also good after, for example, 23.Qxc4 Rac8 24.Qh4 Bc5.
22...Qh2+
Even stronger may be the engines' 22...Bg4!?, eg 23.hxg4 Qh2+ 24.Kf2 (24.Kf1?? Qh1+ 25.Kf2 Nxg4#)  Nxg4+ 25.Ke1 Bg3+ 26.Rxg3 (or 26.Kd1 Qg1+ 27.Qf1 Nxe3+) Qxg3+ 27.Kd1 Nxe3+ 28.Kc1 Nxg2, with a large advantage.
23.Kf1
This is marginally better than 23.Kf2, according to the engines, but Black is anyway winning.
23...Nh5 24.Nce4 Ng3+ 25.Nxg3 Bxg3 26.Qxc4
VK offered a draw, shortly after pressing his clock.
26...Bxb5!
This is best, as it forces mate (or wins the white queen), but other moves win easily enough.
27.Qxb5 Qh1+ 28.Ke2 Qxg2+ 29.Kd1 Qxd2#

Saturday, 14 December 2024

Coulsdon Round One

Spanton (1965 ECF/2012 Fide) - Mukasa Kaddu (1839 ECF/1793 Fide)
Sicilian Bb5(+)
1.e4 c5 2.Nf3 d6 3.Bb5+ Bd7 4.Bxd7+ Qxd7 5.0-0 Nc6 6.c3 Nf6 7.d4!?
A known pawn sacrifice.
7...Nxe4
Stockfish17 reckons immediate acceptance is best, but Dragon1 marginally prefers exchanging on d4 and then taking the e pawn.
8.d5
How should Black proceed?
*****
*****
*****
*****
*****
8...Nb8!?
The mainline in ChessBase's 2025 Mega database runs 8...Ne5 9.Re1 Nxf3+ 10.Qxf3 Nf6 11.c4, when the engines agree White has full compensation for a pawn.
9.Re1 Nf6 10.Bg5 h6
More popular is 10...Na6.
11.Bxf6 gxf6 12.c4?!
Probably better is getting on with development with 12.Nbd2.
12...Rg8!?
This may be a novelty. The engines marginally prefer the known 12...Na6.
13.Qd3
The engines prefer developing the queen's knight to a3 or c3.
13...Qg4 14.g3 Nd7 15.Re4 Qg6 16.Qe2?!
Probably better is 16.Nh4, and if 16...Qg5 then 17.Nd2.
16...Ne5?
Black is at least slightly better after 16...f5, followed by castling, according to the engines.
17.Nh4 Qg5!?
The engines fluctuate between the text and 17...Qh7, in both cases giving White at least a slight edge.
18.f4 Qxh4 19.fxe5 Qg5 20.exd6 f5
How should White proceed?
*****
*****
*****
*****
*****
21.Re5
Not 21.Rf4?? Qxf4, but the engines reckon White has the upper hand after 21.Re3! f4 22.Re4, and if 22...fxg3 then 23.h4!, the point being 23...Qf6? runs into 24.dxe7 Bxe7 25.d6! Qxd6 26.Nc3, with Nd5 and/or Rd1 to come. Possibly better is 22...f5!?, but 23.Re6 is good for White.
21...Kd7!
Best as, although 21...e6 22.dxe6 Bxd6 is fine for Black, White has a major improvement in 22.d7+! Kxd7 23.dxe6+ fxe6 24.Nc3! with a huge attack against the exposed black king.
What should White play?
*****
*****
*****
*****
*****
22.dxe7?
Not taking Black's threats seriously enough. The game is completely equal, according to the engines, after a move such as 22.Qc2 or 22.Qf1.
22...Bxe7?
22...Bg7 wins the exchange.
23.Qf1
Even stronger is developing with 23.Nc3, and if 23...Bf6 then 24.Ne4.
23...Bf6 24.Re2?
Even worse is 24.Qxf5?? Qxf5 25.Rxf5 Bxb2 etc, but winning is 24.Rxf5, eg 24...Qe3+ 25.Kh1 Bxb2 (or 25...Qe4+ 26.Rf3 Bxb2 27.Nd2) 26.Rxf7+ Kd8 27.Nc3!? Bxc3 28.Qf5, with what the engines show is an irresistible attack, one line running 28...Qe8 29.Rf1 Rb8 30.Rf4!? Rg5!? 31.Qh7 Rg8 32.Re4 etc.
24...f4 25.Rg2??
White holds with 25.Kg2 and, maybe, 25.Nd2.
25...fxg3?
Other lines are much stronger, eg 25...Bd4+ 26.Kh1 f3!
26.Rxg3??
After 26.Kh1 Black is a pawn up but the game continues.
26...Qe3+ 27.Kh1
No better is 27.Kg2 Rxg3+ 28.hxg3 Rg8, eg 29.Qf3 Qxf3+ 30.Kxf3 Bxb2 etc.
27...Rxg3 28.hxg3 Qxg3 29.Qxf6
This allows a quick mate, but White's position is anyway hopeless.
29...Qh3+ 30.Kg1 Rg8+ 31.Kf2 Rg2+ 32.Ke1 Qe3+ 33.Kd1 Qe2+ 0-1

Lessons From Benidorm VI: Learn The Basics

AFTER receiving a compulsory bye in round eight, my round-nine game reached the following bishop-and-pawn ending after 30 moves.
The game continued another 24 moves
*****
*****
*****
*****
*****
According to my notes to the game, in those further 24 moves White made five moves deserving a ?, while Black made three such moves, and a further one that received ?!
Admittedly annotations are subjective, but even so the strong implication is that neither player - both rated over 2000 - knew much about what was going on.
LESSON: if your endgame knowledge is barely past beginner level, endings become a lottery.

Friday, 13 December 2024

League Chess

PLAYED on board three (of five) for Battersea against Streatham in Central London League division two last night.

Andrew Stone (2153) - Spanton (1965)
Réti
1.c4 e6 2.Nf3 d5 3.g3 d4!?
This is fifth-most popular in ChessBase's 2025 Mega database, but has been played by grandmasters, including Magnus Carlsen.
4.Bg2 Nc6 5.0-0 Nf6 6.e3 Be7 7.exd4 Nxd4 8.Nc3 0-0
Now both sides have castled, how would you assess the position?
*****
*****
*****
*****
*****
The key to the position seems to be the d pawn. If White can get rid of its backwardness by advancing it to d4, White should be doing well. If Black can keep a piece at d4, or at least replace it with a pawn, Black should at worst have the slight disadvantage that comes with moving second.
9.d3 Re8?
This may be a novelty, and not a good one as it flies in the face of the previous note. Better is 9...Nxf3+ or 9...c5, and even 9...Bc5!?, although the latter involves giving up the bishop-pair after 10.Nxd4 Bxd4 11.Ne2!? (or letting White play 12.d4 after 11...Bb6).
10.Nxd4 Qxd4 11.Be3 Qd8 12.d4 c6 13.Qe2 Qc7 14.Bf4 Bd6 15.Bg5 Be7 16.Rad1 Bd7 17.Rfe1 Rad8
How big is White's advantage?
*****
*****
*****
*****
*****
Black has no weaknesses per se, but White's extra space gives much greater freedom of movement and makes it harder for Black to get counterplay. Stockfish17 and Dragon1 give White the upper hand.
18.Bf4 Bd6 19.Be5!? Bc8 20.Bxf6!?
The engines reckon this is premature, preferring 20.a3, eg 20...a5 21.h4!? f6 22.f4 Qe7 23.g4!?, awarding White the upper hand.
20...gxf6 21.Ne4 Be7 22.Qg4+ Kh8 23.Qh5 f5
The engines suggest 23...Rg8 or 23...Kg7!?
24.Ng5 Bxg5 25.Qxg5 Qe7 26.Qe3 Qf6
Black has beaten off the attack, but at the cost of contracting pawn weaknesses and a very bad bishop
*****
*****
*****
*****
*****
27.Rd2 a6!? 28.Red1 Kg7
This may have been a good time to seek to free the light-square bishop with 28...e5!? After 29.dxe5 Rxd2 30.Qxd2 Qxe5 White is obviously better, but only slightly, according to the engines. Alternatively 29.d5 leads to a different sort of play, but again is probably slightly better for White.
29.f4!
The engines agree that this move, making an ...e5 break difficult to achieve, is best.
29...Re7 30.Qa3 Red7 31.Qc5 h5!?
Striving for kingside counterplay.
How should White respond?
*****
*****
*****
*****
*****
32.h4!?
This is Stockfish17's top choice, and it comes to be Dragon1's, although for quite some time the latter prefers 32.a4. The text prevents ...h4 and fixes the black h pawn on a light square, but is double-edged because it weakens g3.
32...Kh6!? 33.Bf3 Qg7 34.Kf2!? Rg8 35.Rg1 Rgd8 36.Ke3 Rd6 37.Rdd1 R6d7 38.Rg2 Rd6 39.a4 R6d7 40.a5 Rd6 41.Rdd2?!
The engines reckon White keeps the upper hand with 41.Rgg1, and even with 41.Rgd2!?, the point about the latter being 41...Qxg3?? fails to 42.Rg2 Qxh4 43.Qe5 with unstoppable mate.
41...Qf6?!
Missing the chance to more-or-less equalise with 41...f6, when 42...e5 is threatened.
42.Rd3 Qg7 43.Be2 R6d7
The move ...f6 is no longer so effective as the protected state of the white rook on the d file means a quick ...e5 is not threatened.
44.Qe5 Qxe5?!
Probably keeping queens on, as John Carlin (2298) suggested afterwards, is better, although 44...f6 45.Qc5 leaves White with the upper hand, according to the engines.
45.dxe5
The engines slightly prefer 45.fxe5!?
45...Rxd3+?
Much  better is 45...c5, with decent chances to hold.
46.Bxd3 c5 47.Rd2 Rg8 48.Kf2 Rd8
How should White proceed?
*****
*****
*****
*****
*****
49.Rd1
The engines reckon White is winning after protecting the rook with 49.Ke1!? (49.Ke3!? also seems to win), eg 49...Rg8 50.Be2!, when 50...Rxg3?! is met by 51.Rd8, winning the black bishop.
49...Rd4 50.Be2 Rxd1 51.Bxd1 Bd7 52.Ke3 Be8 53.b4!? cxb4?
Black has good drawing chances after 53...Bc6!
The game finished:
54.Kd4 Kg7 55.Kc5 f6 56.Kxb4 fxe5 57.fxe5 f4 58.gxf4 Kg6 59.Bc2+ Kf7 60.Be4 1-0
Battersea defaulted on bottom board, losing the match 5-0.

MY BATTERSEA SEASON 2024-5
DATE.....EVENT...........................COL..RATING..OPP'S RATING..SCORE..SEASON PERF
12/9/24  Central London League   W     1933         1886                   1            2286
26/9/24  Central London League   B      1933         2112                   1            2399
15/10/24 Club Championship        W     1968         1578                   1            2259*
31/10/24 Central London League  B      1968         1867                   0            2061*
19/11/24 London League               W     1957         1991                   1            2127*
21/11/24 Central London League   B      1957         2035                   =           2112*
27/11/24 London League               W     1957         1723                   =           2056*
12/12/24 Central London League   B     1965         2153                   0            2018*
*My season's performance is slightly better than this, the point being my win against the 1578 counts as a 1978 performance, which lowers my average despite me winning the game.

Thursday, 12 December 2024

Lessons From Benidorm V: Similar Is Not The Same

MY round seven game reached the following position.
White has just played 14.Bd3-e2
*****
*****
*****
*****
*****
*****
*****
*****
Stockfish17 and Dragon1 reckon the game is equal after, for example, 14...c6, when most people looking at the Carlsbad pawn-structure might guess the position arose from the Exchange Variation of the Queen's Gambit (it actually arose from a Chigorin, but the structure is the same).
In the Exchange Variation, Black often plays ...Ne4, and that is what I played in the game.
But after 14...Ne4?! the forcing sequence 15.Nxe4 dxe4 (15...Rxe4? 16.Ng5) 16.Ne5 Bxe2 17.Qxe2 gave White the upper hand, according to the engines.
Position after 17.Qxe2
*****
*****
*****
*****
*****
In the first diagram Black has prospects of kingside play.
In the second diagram those prospects have gone, and it is White who is better-placed on the kingside, thanks to the knight on e5 and the opportunity of half-opening the f file.
On top of all that, the e4 pawn proved to be a long-lasting weakness.
LESSON: analogous positions do not always have analogous solutions.

Wednesday, 11 December 2024

Lessons From Benidorm IV: Prelate Power

AFTER taking a half-point bye in round five, my round-six game reached the following position.
Black has just played 19...g7-g5!?
*****
*****
*****
*****
*****
*****
*****
Black has at least a slight edge after 20.Bd2 or 20.Bg3, according to Stockfish17 and Dragon1, but that is probably preferable to the game's 20.hxg5?! fxg5.
The exchange of pawns gives Black's dark-square bishop more space in which to operate, and Black has hopes of creating a passed h pawn (or leaving White with isolanis if an eventual ...g4 is met with fxg4).
White in return gains a protected and passed e pawn, but that is hardly mobile and will not get the support that Black's bishops can give to the black kingside pawns.
The game continued 21.Be3 Be6 22.Kb1 g4, which I met with 23.f4?!
After 23...h4 the following position was reached.
White has a pair of passed pawns, but Black's advanced 2-1 majority is more significant
*****
*****
*****
*****
*****
As I pointed out in my notes to the game, rival pawn-majorities favour a bishop against a knight, and a bishop-pair against a knight and bishop.
This position is unusual in that both the majorities under discussion are on the kingside, but the principle remains the same.
The engines reckon Black has a positionally won game.
LESSON: there are exceptions, eg in a closed position with fixed pawn-chains, but usually a pair of bishops is a powerful weapon.

Tuesday, 10 December 2024

Lessons From Benidorm III: The #1 Factor In Chess

IN round four I reached the following position.
White has just castled
*****
*****
*****
*****
*****
*****
*****
*****
Known continuations are 8...a6 and 8...d6, which are the top choices of Stockfish17 and Dragon1.
I preferred to immediately break White's bind with an apparent-novelty, 8...f5!?
It is not a bad move per se, but Black must be aware after 9.exf5 Nxf5 10.Nxf5 Rxf5 that for the foreseeable future White will have the safer king, which, as grandmaster Nigel Davies enjoys pointing out, is the most important factor in chess.
LESSON: sometimes a move is objectively sound, but results in a player having to walk a thinner tightrope than is desirable for a human, even if engines have little trouble in such positions.

Monday, 9 December 2024

Lessons From Benidorm II: More Trappy Play

AFTER taking a half-point bye in round two, I reached the following position early on in round three.
White has just played 6.Rf1-e1
*****
*****
*****
*****
*****
*****
*****
*****
I played 6...Nh5!?, the tactical point being 7.Nxe5? fails to 7...Qh4 (this is even stronger than 7...Bxf2+), and if 8.Ng4, White has several pleasant choices, the best being 8...d5!?, according to Stockfish17 and Dragon1.
My opponent, a Fide master, not surprisingly avoided capturing the e pawn, and instead played 7.Bf1!?, which is generally desirable, but not the best.
The engines like 7.c3, claiming White is slightly better, despite having played the eccentric 3.Bd3!?
After 7.c3, both 7...Nf4 and 7...Ng3? are met by 8.Bc2, eg 7...Ng3? 8.Bc2 Nh5 (White threatened to win a piece with 9.d4) 9.d4 (anyway!) exd4 10.cxd4, when White has a positionally won game, according to the engines.
LESSON: a trap that works tactically is still bad if the opponent can decline it and emerge better positionally.

Sunday, 8 December 2024

Lessons From Benidorm: Trappy Play

MY round-one game reached the following position after 12 moves.
Black has just played 12...Ng6-f4
*****
*****
*****
*****
*****
*****
*****
*****
What should White do about the g2 pawn - defend it, move it or do something else altogether?
The first option is not very palatable - the king's rook hardly improves its position by moving from behind one pawn on the second rank to behind another pawn on the second rank.
Moving the g pawn seems slightly better, and if 13.g3 is met by 13...Nh3, then 14.Be3 is fine as 14...f5 can be met by 15.Rhe1.
I played 13.Nce2?, after which 13...Nxg2 simply won a pawn as the knight cannot be trapped, eg 14.R(either)g1 is met by 14...Bh3. I continued 14.h4, but Black replied 14...g6, assuring the knight's safety.
However, my basic idea was correct - the g2 pawn is poisoned if White makes the right move, 14.Nde2, as 14...Nxg2? runs into 15.Rxd7! Kxd7 16.Rg1. But 14.Nde2 is not just a trappy way to indirectly defend the g2 pawn - Black is obliged to either retreat the knight or exchange it on e2, so the knight has to leave its forward post, which is why Stockfish17 and Dragon1 reckon 14.Nde2 is the best move in the diagrammed position.
LESSON: when setting a trap, it is vital to set it correctly.

Saturday, 7 December 2024

Winter Plans

HERE are the tournaments I have committed to for the rest of December through to the end of February.

Dec 14-15: Coulsdon Late Autumn in the Surrey town's church-based location. I have entered the five-round open, asking for a half-point bye on Saturday afternoon.

Dec 27 - Jan 4: 35th Cracovia chess festival in Poland's old capital, Kraków. I have entered the 60+ seniors, which is nine rounds over nine days, but that includes a double-round day and a free day on January 1.

Jan 18-25: 23rd Mariánské LáznÄ› chess festival in the former Habsburg spa town of Marienbad. I have entered the 50+ seniors, which is nine rounds over eight days.

Feb 17-26: world senior team championships in Prague. I will be playing for one of the England teams, although I do not know whether that will be in the 50+ or 65+ championships (my priority is to be in a four-person team, rather than a five-person team, the latter probably meaning two byes on top of a free day). Nine rounds over 10 days.

I also hope to play team chess in the London League, the Central London League and the 4NCL.

Friday, 6 December 2024

Summing Up Benidorm

MY score in the open of +0=1-5 (I also took two half-point byes, and was given a compulsory bye) lost 46.2 Fide elo.
A final sunset

Thursday, 5 December 2024

Benidorm Round Nine

Spanton (2009) - Jesús Núñez Muñoz (2060)
English Symmetrical
1.c4 c5 2.Nc3 Nc6 3.g3 g6 4.Bg2 Bg7 5.d3 d6 6.e4 e5 7.Nge2 Nge7 8.0-0 0-0
This position occurs 1,319 times in ChessBase's 2025 Mega database - how would you assess it?
*****
*****
*****
*****
*****
*****
The position is not only symmetrical, but has a closed centre, which greatly reduces the advantage of the move. Stockfish17 and Dragon1 reckon the game is equal.
9.f4 Be6
The mainline in Mega25 runs 9...f5 10.Be3 Nd4 11.Qd2, with an equal position, according to the engines.
10.Nd5 f5 11.Rb1 Qd7 12.a3 fxe4!?
This apparent-novelty - known moves are 12...Nd4, 12...Rab8, 12...Rae8 and 12...Rf7 - is the top choice of the engines.
13.dxe4 Rae8 14.b4 b6 15.Qd2 Nd4 16.Nxd4 cxd4
How would you assess this middlegame position?
*****
*****
*****
*****
*****
Black has a protected passed pawn, but has an extra pawn-island and arguably a less-flexible structure. Dragon1 gives Black a slight edge, but Stockfish17 calls the game equal.
17.Bb2
The engines suggest 17.fxe5!?, eg 17...Rxf1+ 18.Bxf1 dxe5 19.Qd3, when they reckon Black has the better part of equality.
17...Rc8 18.Qd3
The engines again prefer fxe5.
18...Qb7!?
One line suggested by the engines runs 18...b5!? 19.cxb5 Nxd5 20.exd5 Bf5 21.Be4 Bxe4 22.Qxe4 Qxb5, when they reckon Black has the upper hand. Instead 19.fxe5 dxe5 20.Rxf8+ Rxf8 21.Nxe7+ Qxe7 22.cxb5 may be an improvement for White, but Black has more than enough compensation for a pawn after 22...h5, according to the engines.
19.Rbc1 Kh8 20.Rc2
White may be slightly better after 20.a4 or 20.b5!?, according to Stockfish17, but Dragon1 disagrees.
20...b5!?
How should White respond?
*****
*****
*****
*****
*****
21.cxb5?
The engines reckon Black is at best only slightly better after 21.fxe5, eg 21...Rxf1+ 22.Qxf1 dxe5 23.c5 Rf8 24.Qe1 h6!? 25.Nxe7 Qxe7, but there are some tricky points, one being 24...Nxd5 25.exd5 Bxd5 can be met by 26.c6! Qf7 27.Rf2 Qxf2+ 28.Qxf2 Rxf2 29.Kxf2 Be6, when the engines give White complete equality, despite White being a pawn down.
21...Nxd5 22.exd5 Bxd5 23.Rxc8 Rxc8 24.Bxd5 Qxd5 25.Re1 Qc4?
JMM afterwards said he rejected 25...Rc4 because he could not work out what was happening after 26.f5, which he thought was dangerous for Black. However the engines reckon Black is winning, eg 26...Bh6 27.fxg6 hxg6 28.Qxg6? loses to 28...Be3+ 29.Rxe3 (forced) dxe3 as there is no perpetual (the black king can escape to the queenside).
26.Qxc4 Rxc4 27.Kf2?!
Both 27.fxe5 and 27.Rc1 are probably better.
27...Kg8?
Black needed to avoid an exchange of rooks, and this can be done by 27...Rc2+ 28.Re2 Rc4, and by the engines' 27...d3!? 28.Rc1 d2!? 29.Rd1 Rc2.
28.fxe5 dxe5 29.Rc1 Rxc1+ 30.Bxc1 Kf7
How would you assess this bishop-and-pawn ending?
*****
*****
*****
*****
*****
White's farside pawn-majority gives a large advantage - winning, according to the engines.
31.Kf3 Ke6 32.Ke4 Kd6 33.a4 Kc7?
Black can make the win more difficult by taking control of the g5 square with 33...h6.
What is White's winning plan?
*****
*****
*****
*****
*****
34.a5?
The wrong idea - the queenside pawns cannot win by themselves. Correct is 34.Bg5, when the threat is 35.Be7 followed by g4-g5 and Bf6. Black can try to stop this with 34...Kd7, but then 35.a5 wins thanks to the threat of a6 etc. Black could try 34...h6 35.Be7 g5, but 36.g4 leaves the black bishop without a decent move, so the end could come after 36...Kb7 37.Bc5 Kb8 38.a5 Ka8 39.a6 Kb8 40.b6 axb6 41.Bxb6 Bf8 42.Ba5 Ka7 43.b5 Ka8, and now one win pointed out by the engines starts 44.Kd5!? d3 45.b6 e4 46.b7+ Ka7 47.Bc7.
34...Bf6 35.Kd5!? Bg7?!
The engines reckon 35...Kb7 and 35...Be7 are better.
36.g4 Kb7 37.Kc4?
This throws away the win by letting the e pawn advance. Various moves seem to win, including 37.h4!? and 37.Bd2, but the lines are tricky and the engines have a hard time agreeing on what is best.
37...e4 38.a6+
Black to play and draw
*****
*****
*****
*****
*****
38...Kb6
Only 38...Ka8! draws, eg 39.Bb2 d3!? 40.Bc1, when neither side can make progress.
39.Bg5 Kc7
White to play and win
*****
*****
*****
*****
*****
40.Bf4+
Missing a spectacular win with 40.Bd8+!?, when 40...Kxd8? 41.b6 Kc8 42.bxa7 wins trivially. Instead Black can try 40...Kb8, but 41.b6 axb6 42.Bxb6 wins. However, the text also keeps the win alive.
40...Kb6 41.Bg5
Also winning is 41.Bd6 d3 42.Bf4! Be5!? 43.Be3+ Kc7 44.Kd5, when after 44...Bxh2 the engines give as best 45.g5!? Bg1! 46.b6+! axb6 47.Bf4+ Kc8 48.Kxe4 Bh2!? 49.Bd2 b5!? 50.Kxd3 Kc7 51.Ke4 Kb6 52.Bf4! Bg1 52.Be3+ Bxe3 53.Kxe3 Kxa6 54.Kd4 Kb6 55.Kd5 etc.
41...Kc7 42.h3?
But now the win is gone.
42...e3 43.Kc5?
Trying too hard in a completely drawn position.
Black to play and win
*****
*****
*****
*****
*****
43...Be5?
Winning is 43...e2, eg 44.Bd2 Bh6! 45.Be1 (45.g5 Bxg5) d3 etc.
44.Bd8+?
Nice idea ... but much too late. 44.Kc4 draws.
44...Kb8?
Winning is 44...Kxd8, the point being that after 45.b6 axb6+ 46.Kxb6 Black has 46...Bb8.
45.Kc4 Bc7 46.Bf6 Bb6 47.Kd3 Kc7 48.Be7 Kb8 49.g5 Kc8 50.Bd7 Kd7 51.Bc5 Kc7 52.h4 Kc8 53.Bd6 Kd7 54.Bc5 ½–½

Time For A Reset

New moon ... new performance?

Wednesday, 4 December 2024

I Cannot Recall When ...

... I last received a compulsory bye in a non-team event, but it has happened today.
The good news is I get time to explore parts of the Benidorm area that are difficult to get to on foot if playing chess later in the day.
Early-morning sun breaking through cloud cover

Tuesday, 3 December 2024

Benidorm Round Seven

FACED a Spanish junior (born 2009).

Jacobo Villaneuva Bartolome (1884) - Spanton (2009)
Chigorin
1.d4 d5 2.c4 Nc6 3.Nc3 e6!?
More popular are 3...e5, 3...Nf6 and 3...dxc4, but the text has been played by grandmasters, and is liked by Stockfish17 and Dragon1.
How should White respond?
*****
*****
*****
*****
*****
4.cxd5!?
The mainline in ChessBase's 2025 Mega database runs 4.Nf3 Nf6 5.Bg5 Be7 6.e3 0-0 7.Bd3 h6 8.Bh4 b6 9.0-0, with what the engines reckon is an equal position.
4...exd5 5.Bf4 Bd6!? 6.Bxd6
If 6.Nxd5 then 6...Bxf4 7.Nxf4 Qxd4, with equality, according to the engines.
6...Qxd6 7.e3 Nf6 8.Nf3 0-0 9.Bd3 Re8 10.0-0
Now both sides have castled, how would you assess the position?
*****
*****
*****
*****
*****
The pawn-structure is the famous Carlsbad, and the position could easily have arisen from a regular Exchange Variation of the Queen's Gambit, but with the black queen's knight occupying c6 rather than a more normal square for that opening of d7 or f8. The engines reckon the position is equal.
10...Bg4 11.Rc1 a6 12.h3 Bh5 13.a3 Nd8!? 14.Be2
How should Black proceed?
*****
*****
*****
*****
*****
14...Ne4?!
This move is often played in the QGD Exchange, but here it is probably a mistake. Black is fine after 14...c6 or 14...Ne6.
15.Nxe4 dxe4
Not 15...Rxe4? as 16.Ng5 wins the exchange.
16.Ne5
The engines reckon 16.Nd2 is even stronger.
16...Bxe2 17.Qxe2
After a series of forced exchanges, how would you assess the middlegame?
*****
*****
*****
*****
*****
Black no longer has much prospect of pressure against White's kingside. The most significant features of the position are the strongly posted white knight, the advanced but vulnerable black e pawn, and White's potential for play on either flank. The engines give White the upper hand.
17...Ne6 18.Qc2
Best, according to the engines, is 18.f4!? with f5 to come, and if 18...exf3?! then 19.Qxf3 f6 20.Nc4 and 21.Qxb7.
18...f6 19.Ng4!?
The engines prefer 19.Nc4.
19...h5 20.Nh2 Qd5 21.f3 f5 22.fxe4 fxe4 23.Rfe1!? c6 24.Nf1 h4?!
As JVB pointed out after the game, this expansion is weakening rather than strengthening.
25.Nd2 Ng5 26.Nc4 Rad8 27.Qf2 Re6?
The engines suggest 27...c5 or 27...Nf7!?, but much prefer White.
28.Ne5!
This is even stronger than grabbing the h4 pawn, according to the engines.
28...Rh6 29.Qf4 Nh7 30.Rc5 Qe6 31.Ng4 Rg6 32.Re5 Qf7
This is best, according to the engines, but White is winning.
33.Qxf7+ Kxf7 34.Rxe4 Re6 35.Ne5+ Ke7 36.Ng6+ Kd7 37.Rxe6 Kxe6 38.Nxh4
Black has no compensation for being two pawns down
*****
*****
*****
*****
*****
38...g5 39.Nf3 Kd5 40.Rc1 b6 41.Kf2 Rd6 42.Ke2 Ke4?! 43.Nd2+ Kf5 44.Nc4 Rh6 45.Nxb6 g4 46.Rc5+ 1-0