Down to the wire, part I

 "Good endgame technique is essential for any player who wishes to attain the heights of chess mastery. It is only comparatively rarely that games are won in the middlegame and then only as a result of serious mistakes on the part of the opponent. In a significantly greater number of cases it is the endgame which proves to be the scene of a fierce struggle to realise a small advantage built up in the course of the opening and middlegame."

Grigory Levenfish & Vasily Smyslov, Rook Endings (Batsford, 1971)

Those are the very first words in the preface to that celebrated piece of chess literature. But for a long time I doubted if the argument was really true. Admittedly, at my level serious mistakes are commonplace, and I've never questioned the need to study endgames, however middlegame decisions have tended to be the norm in my practice. As I've pointed out in a few other posts though, recently the balance has started to shift the other way.

Whilst I was suffering through the Kingston Invitational back in August, the ECF announced that it was organising a nine-round tournament to be played over a week in Coventry. As the similar events in Cambridge were well-regarded, and had attracted very strong fields, I decided to give it a go. I also resolved to approach this event more thoughtfully than at Kingston. I stayed at the venue for the week, had played some league games beforehand to get in shape, and I used my permitted half-point byes to eliminate two of the three double-round days. Also, bringing us around to the point of this entry, Alekhine stayed on the shelf this time in favour of Capblanca and Karpov.

In the first round I got obliterated by Mark Hebden, but things improved markedly thereafter. The second round was my only double-round day of the event, against a 1910-rated opponent; she chose a bad move order against the Catalan, allowing me to win an exchange and convert it deep in the endgame after a long struggle. After that came this game against Maung Latt (2054). His openings are the stereotypical old man 1. d4 stuff, for which I dipped into the archive to powerful effect. I was soon dominating, but he fought back hard and this game too went down to the wire:

[Event "Coventry International"] [Site "?"] [Date "2025.10.27"] [Round "3.19"] [White "Latt, Maung M"] [Black "Young, Daniel J"] [Result "0-1"] [ECO "D04"] [WhiteElo "2053"] [BlackElo "2017"] [Annotator "Young,Daniel J"] [PlyCount "110"] [EventDate "2025.??.??"] [SourceVersionDate "2025.10.30"] {After a long-overdue win in the previous round, and on the dreaded double-round day no less, now I felt ready to go. My next opponent had a legacy FIDE rating in the 2100s, and with a very toothless opening repertoire this seemed like a real opportunity.} 1. d4 {(1:30)} Nf6 {(1:30)} 2. Nf3 d5 {Rather than my usual King's Indian: I wanted to avoid a London formation, and I spied an opportunity to recycle some preparation from nearly a decade ago.} 3. e3 c5 4. b3 $6 {Here we go! In general the Colle-Zukertort system is playable enough, but against Black's combative move order it doesn't really make sense. The problem is the direct} cxd4 {: White usually recaptures with the pawn to retain some influence over the centre (albeit Nakamura has tried 5. Nxd4 a few times) but he is effectively committed to a queenside fianchetto and the static pawn centre leaves his dark-squared bishop with very poor prospects.} 5. exd4 Nc6 6. Be2 {Departing from my past practice. The bishop is passive here, but placing it on d3 securely requires time that White does not have.} (6. Bb2 g6 {This 'counter-fianchetto' removes any risk of a stock Colle Bxh7+ cheapo, and prepares to up the pressure on the d4 pawn, which is surprisingly vulnerable.} 7. a3 $6 Bg7 8. Bd3 O-O 9. O-O Nh5 10. Re1 $6 {Preserving the d3 bishop from exchange, but now the protection of d4 will require a further loss of time that White can ill afford.} (10. h3 {Stopping ...Bg4 seems to be the best available, but White is left with little to show for giving up the two bishops.} Nf4 11. Re1 Nxd3 12. Qxd3 Bf5 $15) 10... Bg4 11. Be2 {(11. c3? e5! is crushing)} Nf4 12. Nbd2 e5 {Energetic play is logical whilst White's pieces are passively congested. Admittedly, modern engines prefer to hold this in reserve, but White was apparently shell-shocked by my blitzed moves and collapsed with} 13. h3 $2 (13. dxe5 Nxe5 14. Rb1 {isn't mentioned in my old preparation notes, but avoids losing material at once. Still, White faces a cheerless ending after the bonfire sequence} Nxe2+ 15. Qxe2 Nxf3+ 16. Nxf3 Bxb2 17. Rxb2 Re8 18. Qxe8+ Qxe8 19. Rxe8+ Rxe8 20. Rb1 Bxf3 21. gxf3 Re2 $17 {, when his split kingside and passive rook far outweigh the IQP.}) 13... Bxf3 14. Nxf3 e4 15. Nh2 Qg5 $19 {, when g2 and h3 are both attacked, and I went on to win easily in Ismailov-Young, 4NCL 2016. This was all preparation, and one of my proudest achievements in that sphere: in a Najdorf or a King's Indian, being in book on move 15 is child's play, but in a Colle...}) 6... g6 7. O-O Bg7 8. h3 O-O 9. Bb2 Bf5 {Amusingly, I reached an almost identical position in the previous round, only with colours reversed. This time my opponent tries to create some activity on the queenside, but} 10. c4 $6 {(1:25) serves only to loosen his position.} (10. Nbd2 {is more realistic, even if it does amount to playing Black in a dodgy Queen's Indian. After} Rc8 11. c3 a6 $11 {, White's pieces are listless and Black has at least the comfortable side of equality. The queenside fianchetto has clearly not worked out.}) 10... dxc4 $1 $146 {(1:29) I think this is stronger than 10... Qa5 as in the earlier Willin-Schulz, Rottach Egern Seniors 2025. Although Black gives up his pawn centre, White's hanging central pawns prove to be merely a static weakness because his pieces are too passive to make use of the extra space they confer. Black also gains an immediate tempo against the luckless bishop on b2.} 11. bxc4 Qb6 12. Qb3 {In my 2016 notes I only considered the silicon move 12. Qc1, but I was older and wiser when I reviewed them for this game, and knew that a human would want to dampen things down. I was still in preparation.} Ne4 13. Qxb6 axb6 $17 {It is counterintuitive, but the exchange of queens has helped Black. White is deprived of one of his few well-placed pieces, as well as a protector of d4, and a new target appears on a2 thanks to the opening of the a-file. As for the doubled b-pawns, per Nimzowitsch they are only a weakness if they can be attacked.} 14. g4 $6 {This is a lunge to be sure, but not as random as it looks since Black hasn't got a perfect retreat: now I have to choose between blocking the d-file and therefore my pressure on White's pawns, disconnecting my rooks, or worrying about d4-d5 forks. But objectively this is not a good move, for had I seen through the mirage then only the new pawn weakness would have been left behind.} Bc8 $6 {Not exactly bad, but lazy. Black would rather keep his rooks talking, and 14... Be6!, not fearing the d-pawn, would have allowed me to do this without compromising my assault on d4.} (14... Be6 $142 $1 {Of course, White can't play 15. d5 at once because of the hanging pieces on the long diagonal, but I did in fact notice this. What was harder to grasp was the time factor: since there's no easy way to defend the bishop on b2, d4-d5 will not be a threat any time soon, and my piece play will bear fruit long before then. I made an almost identical misjudgement in Wong-Young, Wallington Club Ch. 2023.} 15. Rd1 f5 {As well as chewing up White's kingside pawns, this also secures the knight on e4 and vacates f7 for my light-squared bishop. After} 16. g5 Ra4 17. Nbd2 Rd8 $17 {White's pawn centre is attacked from all directions.}) 15. Kg2 $2 {(1:13) At the time I was baffled by this move, but perhaps White (justly) feared that his h-pawn could come under fire with a later ...h5. Still, the objection to this move is obvious: his centre is collapsing and he has no time for such niceties. The natural 15. Rd1 would have been stronger.} Rd8 {(1:23)} 16. Rd1 $6 {Moreover, this necessary defensive move now has an important tactical hole...} Nc5 $6 {... which I didn't see. The text move was an idea transplanted from elsewhere in my preparation; the looming threat of ...Na4 spells the end for White's efforts to hold onto his d-pawn. However, Black can do even better.} (16... h5 $142 {It turns out that White's 15th move was in vain, since} 17. gxh5 {runs into} (17. Nc3 $142 {, belatedly developing as in the game, seems best, but} hxg4 18. Nxe4 gxf3+ 19. Bxf3 Bxd4 $19 {wins a pawn with none of the lingering activity that White gets after my move}) 17... Bxh3+ $1 18. Kxh3 Nxf2+ {and now we see why the rook is unhappy on d1.} 19. Kg2 Nxd1 20. Bxd1 Nxd4 $19 {Black already has a rook and two pawns for the two pieces, but White's troubles are not over yet since the tactical weakness of the long diagonal will cost him his h-pawn too. Note that if} 21. Nxd4 {Black should continue} Rxd4 $1 {, maintaining the agony.}) 17. Nc3 {Undoubtedly the best try. Rather that contort himself any further, White lets the d-pawn go and tries to activate his pieces.} Nxd4 18. Nb5 {If my rooks were connected (cf. move 14) this would be a dead end for White, but as it is he gets a chance to disoraganise my pieces a little.} Nc6 19. Rxd8+ Nxd8 20. Nc7 {(41)} Ra5 {(1:07)} (20... Bxb2 $143 $4 {fails to} 21. Rd1 $16 {when two pieces are attacked - at the risk of labouring the point, Black really wishes his light-squared bishop was somewhere else right now! I didn't look further than this, and for good reason as White may even be winning:} Rb8 $2 22. Rxd8+ Kg7 23. Ne8+ Kh6 24. h4 $1 $18 {and out of nowhere Black has been caught in a mating net.}) 21. Bxg7 Kxg7 22. Nd5 {Realising that I couldn't keep my extra pawn, I decided that it was now my turn to quietly improve the pieces. Black retains some positional advantage owing to White's split queenside pawns and his bad bishop, which is hemmed in by the c- and g-pawns. Having said that, the method I selected was not the best.} Nc6 $6 {It seems sensible not to block my bishop's path back into play, but routing this knight to e6 forces White to keep an eye on f4 and so limits his options.} (22... Nde6 $142 {Whilst this only delays the loss of the pawn, it does get a backwards step out of White:} 23. Bf1 (23. Nxb6 $143 $2 {(or equally 23. Nxe7) is met by} Nf4+ 24. Kf1 Be6 $19 {, when White will shortly lose a pawn either on c4 or h3 - Black's knights are terribly strong.}) 23... Kf8 24. Nxb6 Bd7 {Unlike in the game continuation, Black keeps a4 under control. White can, and should, play} 25. Nxd7+ {, as otherwise the bishop will be too powerful on the long diagonal, but after} Nxd7 $17 {Black's pieces are better as well as his structure, and the a-pawn is restrained.}) 23. Nxb6 Be6 24. Bf1 $6 {Though it was a necessary precaution in the line above, this retreat makes an odd impression here - now I am exonerated for my imprecise 22nd move. 24. a4, safeguarding the a-pawn and denying Black a route to the third rank, was more stubborn.} Ra3 25. Nd5 {(27) Having fixed the a-pawn as a weakness, now I needed a follow-up. I opted for a tactical approach having seen that the tempting 25... Nb3 doesn't achieve anything after 26. Rb1, but Black can refine this idea...} b5 $6 {(51) Similarly to my 14th and 16th moves, this keeps most of my advantage but isn't the most incisive. Black must be on guard against cashing in prematurely.} (25... f5 $142 {A theme that Karolyi brings up in his study of Karpov's games is how often he wins the strategic battle on 'his' flank and then proceeds to do the same on the other wing - he aims to dominate the whole board. This could be an example from the Karpovian playbook: with White nursing his weak queenside pawns, Black starts making inroads against the kingside structure, and ultimately uses this to break in on a2 after all.} 26. gxf5 (26. g5 $144 {allows Black to anchor his c5 knight with} Ne4 {, and ...Bxd5 followed by ...Nb4 becomes a nasty threat since ...Rxa2 will also hit f2. White lacks a good defence to this idea, for instance} 27. Nf4 Bf7 $19 {and now ...Nc3 comes into the equation.}) 26... Bxf5 {Here Black can't settle a knight on e4 so easily, but Black's bishop comes to life on its new diagonal. Crucially,} 27. Ne3 Be4 28. Kg3 {allows Black to win a pawn by} Bxf3 $1 29. Kxf3 Nb3 {, and White can't save himself with Rb1 any more because of the fork on d2.} 30. Re1 Rxa2 $17) 26. Ne3 $6 {When I first saw this move I thought - not for the first time in this game - that it had to be too slow, but on this occasion things are just about holding together for White. However, more active defence was possible.} (26. Rd1 $142 $1 {As before, activity is more important than sickly pawns. Black may as well duly win the a-pawn with} bxc4 27. Bxc4 Na5 28. Bb5 Rxa2 {, however} 29. Nc7 Nab7 $15 {leaves Black's knights in a twist and offers White some drawing chances, especially as all of the pawns are on one side.}) 26... b4 27. Nc2 $8 {Forced, otherwise 27... b3 is terminal. Now White will answer 27... Rc3 with the insouciant 28. Ne3.} Ra4 {I chose to maintain the attack on a2, but White can secure b3 with} 28. Nd2 {. Now I wasn't sure how to continue, as no forcing measures against the weak queenside pawns are apparent. After some thought, I did something unusual for me,} g5 $5 {- I made a purely strategic move. I wasn't trying to make quick gains now and instead opted to prepare for a future endgame by fixing White's kingside pawns on the same colour as the bishops.} 29. f3 h6 30. Kf2 {(26)} Ra8 $6 {(37) Making to redeploy my rook to the d-file, but given I felt it best to exchange rooks after White's reply, this is a wasted tempo.} (30... f5 $142 {The threat of a kingside clamp with ...f4 drives White to exchange without waiting for Black's rook to move.} 31. Nb3 (31. gxf5 $2 {won't do because} Bxf5 {wrests control of c2, so} 32. Ne3 Bg6 $19 {and now ...b3 is an unstoppable threat.}) 31... Nxb3 32. axb3 Rxa1 33. Nxa1 f4 $17 {Black is making quicker progress than in the game.}) 31. Nb3 (31. Ke3 $143 $2 {White can't delay the text move any further. Bringing the king in first gives Black time to establish the bind with} f5 {, when} 32. Nb3 (32. Be2 {If White waits then} f4+ 33. Kf2 Bg8 $1 {decisively re-routes to the b1-h7 diagonal, after which White's blockade of b3 is broken.}) 32... Nxb3 33. axb3 Rxa1 34. Nxa1 f4+ 35. Kd2 Nd4 $19 {will win either the b-pawn or the f-pawn: White's king cannot protect them both.}) 31... Nxb3 32. axb3 Rxa1 33. Nxa1 {Even though the computer is now (barely) able to defend White's cause, in practice things are far from easy - I have succeeded in locking his pawns on the wrong squares for his bishop, and both of his minor pieces are passive. Next, I centralise my king.} Kf6 34. Ke3 Ke5 35. Nc2 {(25) White counters one idea, not letting me establish my knight on d4, so I go over to another and gain some space with the f-pawn.} f5 {(26)} 36. Be2 $6 {Sitting tight is the right approach, but it would have been more natural to place the bishop actively on d3. The key point is that White gets the opportunity to exchange off his bad bishop.} (36. Bd3 $142 f4+ 37. Kd2 Kd6 {Heading for c5. As already mentioned, White's defensive trump is} (37... Nd4 {As in the game, the exchange of b-pawns eases White's defence, since his king is closer to the remaining queenside weakness on c4:} 38. Nxb4 Nxb3+ 39. Kc3 Nc5 40. Be2 $11) 38. Be4 {, a trade that Black isn't in time to avoid because of the attack on his b-pawn.} Bd7 39. Bxc6 Bxc6 40. Ke2 Kc5 {Black is on the point of putting his bishop to work on those light-squared kingside pawns with ...e5-e4, but White is just in time to save himself with} 41. Ne1 $8 Kd4 42. Kd2 e5 43. Nc2+ Kc5 44. Ne1 $11 {, and Black can't approach any closer. If} e4 45. fxe4 Bxe4 46. Nd3+ Kd4 $5 {(the pawn ending is also a draw)} 47. Nxb4 Bg2 48. Ke2 $8 Bxh3 49. Kf3 Bf1 50. Nc6+ Kc3 51. c5 Kxb3 52. Ne7 {Black will have to give up his bishop to stop the c-pawn.}) 36... f4+ 37. Kd2 Nd4 $6 {This is the only point where my directness hurt me. I reasoned that isolating the c-pawn would make it easier to attack, however this is offset by White recovering the c3 square. In general, pawn exchanges help the defender.} (37... Kd6 $142 {This time, if White tries to exchange off his troublesome bishop with} 38. Bd3 {, Black is in time to defend b4 and then evade it with} Kc5 39. Be4 Na5 40. Na1 Bf7 41. Kc2 h5 $15 {I'm not sure Black can force his way in even here, but at least there's no easy escape for White.}) 38. Nxb4 Nxb3+ 39. Kc3 Nc5 40. Nc6+ {(23)} Kd6 {(22)} 41. Nd4 Bd7 42. Nc2 $6 {This is the start of a losing plan. White is steering for d5, but this does nothing to stop Black putting his pieces on optimal squares. The computer finds a more active defence.} (42. Kb4 $142 {Now that Black's knight is attacked, his king loses some flexibility. The obvious reply is} e5 {, but here the resourceful box is ready with} 43. Nf5+ $1 Bxf5 44. gxf5 {, and the separated passed pawns give White enough counterplay to hold in spite of his apppalling structure - now that it has no opponent, the bishop can show its advantages over the knight.} Nd7 45. Kb5 h5 46. Bd1 Nf6 47. c5+ Kc7 48. Kc4 Kc6 49. Ba4+ Kc7 50. Kd3 $11 {Black can't approach the c-pawn, his knight must maintain its vigil over e4 and g4, and so there is no progress to be made.}) 42... e5 43. Nb4 Ne6 44. Nd5 $2 {The losing move, and there will be no reprieve this time. As impressive as d5 looks for White's knight, it forfeits his defence of f3.} (44. Nd3 $142 {still holds; Black's position looks overwhelming after} Nd4 45. Bd1 Bc6 46. Ne1 Kc5 {, but White narrowly avoids sliding into zugzwang with} 47. Nd3+ $8 Kd6 48. Ne1 $11) 44... Nd4 45. Kd3 {(9)} Bc6 {(16) White's king is overloaded by the defence of both minor pieces, and a desperate counterattack is the only hope left for him. But Black's passed pawns will sweep away all obstacles.} 46. Nf6 Nxf3 {I spent some time trying to figure out which piece to capture with (and, implicitly, to have traded off), and chose correctly. 46... Bxf3?! also appears to win, but keeping my bishop promises me White's remaining kingside pawns as an insurance policy and provides an escort for the e-pawn.} 47. Bxf3 Bxf3 48. Ng8 e4+ 49. Kd4 e3 50. Kd3 {(5)} Bg2 {(5)} 51. Ke2 (51. Nxh6 $144 {If White follows through on taking this pawn then} Bf1+ 52. Kc3 Ke5 {leads to the same result; on} 53. Nf7+ Kf6 54. Nd6 f3 $19 {and wins.}) 51... Ke5 {Black can afford to ignore the c-pawn, which is much too far back. Now his king reaches e4 and the rest is easy.} 52. Nxh6 Ke4 53. Nf5 f3+ 54. Ke1 f2+ 55. Ke2 {(0) My opponent was now on increment and very taken aback by} f1=Q# {(5), capping off a game that I had to win twice over. Having largely wasted some good preparation, it was satisfying to grind out a position I thought had gotten away from me - even if it did need a little help from my opponent.} 0-1

Whilst I might well have won this game a lot sooner, my hard-fought victory in the endgame was well deserved even if Latt did heroically fight his way back to equality. One challenge of slavishly following engines is that an emotionless number doesn't tell you how difficult a position is. Even if you accept that a 0.00 evaluation is objectively correct, the method may not be obvious and there's no guarantee of a margin for error. If you put an engine on the position after 33. Nxa1, it gives an evaluation of about -0.10. Strictly speaking that is correct - Black has a token advantage in an equal position - but it belies the challenge confronting White after a tiring defence with time trouble on the horizon. His knight is in the corner, his bishop is boxed in by its own pawns, and Black can press without taking risks. Tellingly, White didn't lose because of a terrible error, but because he let me put my pieces on natural squares and his position couldn't cope with the pressure. In the end it was a lesson to keep on playing, because I'd known for a long time that I had wasted my opening advantage. The reader may also note that I avoided the wild excess of my games at Kingston, and kept things simple. In the game there against Cowan I irrationally refused a queen trade and nearly paid for it; here I willingly provoked one.

In round 4 I was drawn into complications in a sharp Ruy Lopez and got outplayed by yet another 2300+ opponent, before taking a bye that afternoon. The next day I landed an even cleaner hit with preparation and got a big advantage against a 2258-rated junior, but simplified too quickly and decided to settle for a draw in a slightly better position to conserve energy: I was getting ill by the end of the tournament. After stumbling into an unpleasant N vs B endgame it was my turn to get ground down from objective equality in round 7, and I spent my round 8 bye bedridden. But I was able to end on a positive note with a preparation knockout in the last round, winning in only 23 moves - the first time I've won in the last round of a major tournament since 2019! That gave me a 50% score and a 2122 TPR, a clear improvement over Kingston and perhaps proof that what doesn't kill you makes you stronger. Playing one game a day definitely helped, as I did a lot of damage in the opening, and playing some quieter games was a necessary correction.

I could linger on this event and go over some other games, but on reflection this was representative in itself. Therefore we'll be going back to the Oxfordshire League for the next game, which was another on the same theme.