The joy of endgames

by Jonathan Speelman
11/7/2021 – All chess analysis, and that of endgames in particular, has been transformed in recent years by computer technology. Not only can you turn an engine on to get an instant assessment, which is generally fairly correct, but when you get down to seven pieces, then you have the ear of God — or rather an omniscient tablebase. In today’s column, we look at two fascinating endgames from the Grand Swiss in Riga. | Photo: Mark Livshitz

ChessBase 17 - Mega package - Edition 2024 ChessBase 17 - Mega package - Edition 2024

It is the program of choice for anyone who loves the game and wants to know more about it. Start your personal success story with ChessBase and enjoy the game even more.


Two endgames from Riga

[Note that Jon Speelman also looks at the content of the article in video format, here embedded at the end of the article.]

FIDE Grand Swiss 2021Last time, I looked at some games by the world’s top juniors, and I was thinking of adding some more now. Indeed, one of today’s games is by a junior, but since he’s a top ten player already, Alireza Firouza’s win against Evgenij Najer hardly counts as a triumph for sprogdom  (or is it -hood?).

It comes from the FIDE Grand Swiss in Riga (where, as I wrote this sentence, Firouzja resigned against Fabiano Caruana, leaving a three-way tie between the two of them and David Howell with two rounds to go). And I’ve chosen it because of the interesting endgame which he managed to win.

Today’s other game is another ending from Riga — the pawn endgame which Maxime Vachier-Lagrave abandoned as a draw against Alexei Shirov without apparently really believing that he had serious chances.

All chess analysis, and that of endgames in particular, has been transformed in recent years by computer technology. Not only can you turn an engine on to get an instant assessment, which is generally fairly correct (though can be skewed by successful fortresses for example), but when you get down to seven pieces, then you have the ear of God — or rather an omniscient tablebase. This makes it all hugely easier than hitherto, and I trust that I’ve got a reasonable account of both endings in not much more than an hour.

We start chronologically, with the rook ending.

[The games are presented both in plain text (with diagrams), and on our dynamic replayer board at the end of the article.]

Firouzja, Alireza (2770) - Najer, Evgeniy (2654) 
FIDE Grand Swiss 2021 (7.1), 03.11.2021
[Speelman, Jonathan]

1.e4 e5 2.Nf3 Nf6 3.d4 Nxe4 4.dxe5 d5 5.Nbd2 Nxd2 6.Bxd2 Be7 7.c3 c5 8.Bd3 Nc6 9.0-0 Bg4 10.Re1 Qd7 11.h3 Bh5 12.Bf4 Qe6 13.Be2 0-0 14.Qd2 Bg6 15.Rad1 Be4 16.Ng5 Bxg5 17.Bxg5 d4 18.Bf1 Qg6 19.Qf4 Bc2 20.Rd2 Rae8 21.Bh4 Kh8 22.f3 a6 23.Bf2 Rd8 24.Rc1 Bb1 25.cxd4 Nxd4 26.Bxd4 cxd4 27.a3 Ba2 28.Rxd4 Qb6 29.Rcd1 Rfe8 30.Kh2 Rxd4 31.Qxd4 Qxd4 32.Rxd4 g5 33.Rd7 b5 34.Rd6 Rxe5 35.Rxa6 Bc4 36.Bxc4 bxc4 37.a4 Rc5 38.Rb6 c3 39.bxc3 Rxc3 40.a5 Kg7 41.a6 Ra3 42.Kg3 Ra2 


With his king boxed in, Firouzja had to sacrifice a pawn to enable him to move towards the queenside.

43.f4 gxf4+ 44.Kxf4 Rxg2 45.Ke5 Ra2

[Since Ra2 should draw there was no need to look further, but I also wondered about 45...Rh2 which turns out to lose: 46.Kd6 Rxh3 47.Kc7 Ra3 48.Kb7 h5 49.a7 h4 (49...Rxa7+ 50.Kxa7 h4 51.Rb4 f5 52.Rxh4 Kf6 53.Kb6 Ke5 (53...Kg5 54.Rh8 f4 55.Kc5) 54.Kc5 f4 55.Kc4 Ke4 56.Kc3 Ke3 57.Rh8 f3 


Analysis arising from 45...Rh2 and 49...Rxa7+.

This is a critical position in rook v pawn which you can calculate beckwards from. With the kings and pawn like this on any file (a rook’s pawn is slightly different, but the result is the same), White to play wins, while Black to play draws.

58.Re8+ Kf2 59.Kd2 Kg2 60.Ke3 f2 61.Rg8+ Kf1 62.Rf8) 50.Ra6 


50...Rb3+ 51.Kc7 Rc3+ 52.Kd6 Rc8 53.a8Q Rxa8 54.Rxa8 Kf6 55.Rh8 Kg5 56.Ke5 Kg4 57.Ke4 Kg3 58.Ke3 h3 59.Rg8+ Kh2 60.Kf2 f5 61.Rh8 f4 62.Rh7 f3 63.Rh8 Kh1 64.Rxh3#]

46.Kd6 f5 47.Kc7 f4 48.Kb8 f3 49.a7 f2 50.Rb1 


Game critical position. In this very tense position, Black could have drawn with either Kf6 or, a little surprisingly, Rb2+ — but chose a third losing option. 50...f1Q?

[50...Rb2+! 51.Rxb2 f1Q

And here rather than drown in sequences of checks you can simply ask a tablebase, which goes ping and tells you that it’s a draw:

52.Rb7+ (52.a8Q Of course you don’t even need an engine to verify this one: 52...Qf8+ 53.Kb7 (53.Ka7 Qa3+) 53...Qf3+ 54.Kb8 Qf8+) 52...Kg6 53.Rb6+ (53.a8Q Qf8+ 54.Ka7 Qa3+) 53...Kg7 Kg5 and Kh5 also hold 54.a8Q Qf8+ 55.Ka7 (55.Kb7 Qe7+ 56.Ka6 Qa3+) 55...Qa3+ 56.Ra6 Qc5+ 57.Kb7 Qe7+ 58.Kb6 


And now Qb4+ and Qe3+ are the two moves to draw: 58...Qb4+ Okay, we’ll believe the tablebase — and indeed it does look very plausible that White can’t escape the checks (58...Qd6+?? 59.Qc6); 

50...Kf6! In the diagram (game  critical position) Kf6! is the cleanest way to draw, though at this moment it’s visually a bit surprising (to me at least) that the pawn v rook ending holds by a tempo.

51.a8Q Rxa8+ 52.Kxa8 Kg5 53.Rf1 Kh4 54.Rxf2 Kxh3= 55.Rf6 h5 56.Rg6 h4 57.Kb7 Kh2 58.Kc6 h3 59.Kd5 Kh1 60.Ke4 h2]

51.Rxf1 Rb2+ 52.Ka8 Rb3 53.Rc1 Kg6 54.Rc7 h5 55.Rb7 Rxh3 56.Rb6+! The cleanest.

[56.Rb5 Was another thematic way to win, cutting the king off from the rook's pawn's advance. 56...Ra3 (56...Rf3 57.Kb7 Rf7+ 58.Ka6 Rf8 59.Rb8 Rf1 60.a8Q Ra1+ 61.Kb5 Rxa8 62.Rxa8) 57.Kb7 h4 58.a8Q Rxa8 59.Kxa8 h3 60.Rb3 h2 61.Rh3] 56...Kg5 57.Kb7 Ra3 58.Ra6 Rb3+ 59.Kc7 


Black is unable even to give the rook up for the pawn: so he resigned. An excellent ending by Alireza.

[59.Kc7 Rc3+ 60.Kd7 Rd3+ 61.Ke7 Re3+ 62.Kf7 Rf3+ 63.Kg7] 1-0

Evgeniy Najer, Alireza Firouzja

A long struggle — Alireza Firouzja and Evgeniy Najer | Photo: Anna Shtourman

Vachier-Lagrave, Maxime (2763) - Shirov, Alexei (2659)
FIDE Grand Swiss 2021 (8.2), 04.11.2021
[Speelman, Jonathan]

1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 b5 6.Bb3 Bc5 7.a4 Rb8 8.c3 d6 9.d4 Bb6 10.a5 Ba7 11.h3 Bb7 12.Be3 Nxe4 13.Nbd2 exd4 14.cxd4 Nxd2 15.Qxd2 0-0 16.Rfe1 Qd7 17.d5 Bxe3 18.Qxe3 Ne5 19.Nxe5 dxe5 20.Qxe5 Rbd8 21.Rac1 Rfe8 22.Qd4 Rxe1+ 23.Rxe1 Qd6 24.Qe3 h6 25.Rd1 Bc8 26.Rc1 Bd7 27.Qg3 Qxg3 28.fxg3 Rc8 29.Rf1 Kf8 30.d6 f6 31.Rd1 cxd6 32.Rxd6 Rc1+ 33.Kf2 Bc6 34.Re6 b4 35.Ke3 Bb5 36.Kd4 Ra1 37.Kc5 Rxa5 38.Rb6 Ke7 39.Rb7+ Bd7+ 40.Kxb4 Re5 41.Ba4 Re4+ 42.Kb3 Rxa4 43.Rxd7+ Kxd7 44.Kxa4 Kc6 45.Ka5 Kb7 


Here I think that MV-L had almost given up convinced that it was drawn, and indeed after a couple more moves, over which admittedly he took about 20 minutes, they agreed the draw. I also thought that because the a6-pawn controls b5 it must be drawn. But I was watching with a pupil who very perceptively demurred suggesting that since White’s king is better he has  chances, and indeed it’s absolutely fascinating, though I believe that the draw is the “correct” result.

46.g4 Ka7 47.h4 Kb7 



[48.h5? Now it’s dead becasuse even if White gets his king to g6 he will have no break on the kingside, so Black can simply allow Kf5-6 and meet this with Kf8. Here they agreed the draw. I’ve promoted Kb4 so that the main line appears in bold.]


[I believe that Kc6 should draw, though White can reach queen and  h-pawn v queen. 48...Kc6 49.Kc4 Kd6 50.Kd4 g6! (50...Kc6 51.Ke4 Kd6 52.Kf5 Ke7 53.Kg6 Kf8 54.g5+-) 51.Ke4 Ke6 52.g3 


(52.h5 gxh5 53.gxh5 f5+ 54.Kd4 Kf6) 52...f5+! 53.gxf5+ gxf5+ 54.Kd4 Kd6 55.b3 (55.h5 a5 56.b3 comes to the same thing.; 55.b4 h5!=) 55...a5 56.Kc4 Kc6 57.h5 Kb6 58.Kd5 Kb5 59.Ke5 Kb4 60.Kxf5 Kxb3 61.g4 a4 62.g5 a3 63.g6 a2 64.g7 a1Q 65.g8Q+ 


End of the line starting 48.Kb4 Kb6. You don't really need a tablebase to know that this ought to be drawn, but of course in practice  White would fight on.]

Returning to the line where Black plays Kb4-Kb6.

49.Kc4 Kc6

[49...a5 50.h5 Kc6 51.g3 Kb6 52.Kd5 Kb5 53.Ke6 Kb4 54.Kf7 Kb3 55.Kxg7 Kxb2 56.g5 fxg5 57.Kxh6 a4 58.Kxg5 a3 59.h6 a2 60.h7 a1Q 61.h8B+]

50.g3 To induce Kd6.

[If 50.Kd4 Kb5 51.Kd5 Kb4 52.Ke6 Kb3 53.Kf7 Kxb2 54.Kxg7 a5 55.Kxf6 a4 56.g5 hxg5 57.hxg5 a3 58.g6 a2 59.g7 a1Q 60.g8Q Qf1+ 


White tries to avoid 50 g3. Black is able to counterattack, and this draws pretty easily.]


[I wondered  about a5 to try to race but in fact this this is totally lost: 50...a5? 51.h5 Kb6 (51...a4 52.Kb4) 52.Kd5 Kb5 53.Ke6 Kb4 54.Kf7 Kb3 55.Kxg7 Kxb2 56.g5! fxg5 57.Kxh6 a4 58.Kxg5+-]



51...Kc6? This seems to lose to a very exact sequence.

[51...g6 This leads to a queen endgame which is “drawn”, but would be very unpleasant to defend in practice.

52.h5 g5 53.b4 (53.Ke4 Ke6 54.b4 Ke7!= 55.Kd5 (55.Kf5 Kf7) 55...Kd7) 53...Kc6 54.Ke4 Kb5 55.Kf5 Kxb4 56.Kg6 a5 57.Kxh6 a4 58.Kg6 a3 59.h6 a2 60.h7 a1Q 61.h8Q

You can’t use a tablebase here because there are eight pieces, but after a few sample variations in which a pawn came off, I was then able to check, and I’m pretty confident that this is a draw, e.g.:

61...Qb1+ 62.Kg7 Qe4 63.Qc8 Qe5 64.Qf5 Qd6 


Line starting 51...g6. Black can draw but must work  for it. 65.Kg6

(65.Qxf6 Qxg3 is also a draw.)

65...Ka3 The king normally gets away from the passed pawn unless it can get in front. This  minimises the danger of cross checks later. 66.Qxf6 Qxg3]

52.Ke4 Kd6 53.Kf5 Ke7 54.Kg6 Kf8 55.g5 fxg5 56.hxg5 hxg5 


Here I automatically recaptured Kxg5, but then wondered whether White could improve with g4 first, and the  tablebase tells me that this is winning!

[56...h5 57.Kh7 Kf7 58.g6+ Kf6 59.b4]




This critical position reached after White recaptures 57.Kxg5? is drawn. I’d imagined that would be winning but in fact tablebases tell us that it’s a draw. And when you investigate you realise that this is partly because Black can sometimes run to the queenside and get king and pawn a2 v queen; and partly because he can sometimes go and take the g-pawn while White is taking his and then get in Kc4 after White plays Kxa6. I should mention that it’s drawn only because White had to play g2-g3. With the pawn on g2 this is winning with either side to move.


(57...Kf7 also draws 58.Kf5 Ke7 59.g4 (59.Kg6 Ke6!) 59...a5! 60.Kg6 Kf8 and 61.Kh7? now loses because the tempi have changed on the queenside. a) 61.g5 Kg8 62.Kf5 Kf7 63.g6+ Ke7 64.b3 (64.Ke5? a4-+) 64...Kd6 65.Ke4; b) 61.b3 Kg8 62.Kg5 Kf8 63.Kf5 Kf7; 61...Kf7 62.g5 a4-+)



(58.Kg6 Ke6! 59.Kxg7 (59.g4 Ke5 60.Kxg7 Kf4 61.Kf6 Kxg4 62.Ke5 Kf3 63.Kd4 Ke2 64.Kc5 Kd3 65.Kb6 Kc4 66.Kxa6 Kb3)

59...Kf5 This looks too slow, but Black draws by a tempo.

60.Kf7 Kg4 61.Ke6 Kxg3 62.Kd6 Kf3 63.Kc6 Ke4 64.Kb6 Kd4 65.Kxa6 Kc4




(58...Kf7? 59.b3 (59.b4? Ke7 60.Ke5 Kd7 61.Kd5 Kc7) 59...Ke7 


60.g4! By advancing the g-pawn first, White prevents Black from running to the queenside and racing.

a) 60.Ke5 Kd7 61.Kd5 Kc7 62.g4 (62.Ke6 Kb6) 62...Kb6!;

b) 60.b4? Kd6; 60...Kf7 61.b4! 


Zugzwang. White to move can only draw since Black keeps the opposition.

61...Ke7 62.Kg6 (62.g5 Kf7 63.g6+ Ke7 64.Ke5 Kd7 65.Kd5 Ke7 66.Kc6) 62...Kf8 63.Kh7 Kf7 64.g5 White has rectified the tempi on the queenside and now wins since it’s absolute zugzwang (White to move would lose).)

59.Kg6 Kc5 60.Kxg7 Kb4 61.g4 Kb3 62.g5 Kxb2 63.Kf6 a5 64.g6 a4 65.g7 a3 66.g8Q a2= 





After 57 g4! a5. And now the  tablebase tells me that Kf5 is the only winning move!

[57...Kg8 58.b4 Kf8 59.Kxg5 Kf7 forced because there are no tempi left on the queenside. (59...Ke7 60.Kg6 Kf8 61.Kh7 Kf7 62.g5; 59...Kg8 60.Kg6 Kh8 61.Kf7 Kh7 62.g5 Kh8 63.g6 a5 64.b5) 60.Kf5 Ke7 61.Kg6 Kf8 62.Kh7 and wins.]


[58.Kxg5 a4 59.Kg6 Kg8 60.g5 Kf8 61.Kf5 Kf7 62.g6+ Ke7 63.Ke5 Kd7 64.Kd5 Ke7 65.Kc5 and they queen at the same time.; 

58.b3? Here the pawn is one square closer to the black king, which means that he can run to the queenside. 58...Ke7 59.Kxg7 Kd6 60.Kf6 Kc5 61.Kxg5 Kb4 62.Kh6 Kxb3 63.g5 a4 64.g6 a3 65.g7 a2 66.g8Q+ Kb2=]

58...Ke7 59.Ke5 


Because of the g5-pawn, Black can’t reach the g4-pawn in time and White is able to win the race.


[59...Kd7 60.Kd5]

60.Kd5 Kf6 61.Kd6! White can wait until the black king retreats and then win by a tempo 61...g6 62.Kd5 a4 63.Kd6 Kf7 64.Kc5 Ke6 65.Kb4 Ke5 66.Kxa4 Kf4 67.b4 Kxg4 68.b5 Kh3 69.b6 g4 70.b7 g3 71.b8Q g2 72.Qb6 Kh2 


The extra g-pawn complicates matters slightly, but as long as White is careful he will win.

73.Qf2 g5 74.Kb3 Kh1 75.Qf3 g4 76.Qe4 Kh2 77.Qf4+ Kh1

[Or 77...Kh3 78.Qf2 Kh2 79.Qh4+]

78.Qh6+ Kg1 79.Kc2 g3

[79...Kf1 80.Qf4+ Ke2 81.Qxg4+]

80.Kd2 Kf1 81.Qf4+

[Or indeed 81.Qe3 g1Q 82.Qe2#] ½-½


Select an entry from the list to switch between games

Magical Chess Endgames Vol. 1 & 2 + The magic of chess tactics

In over 4 hours in front of the camera, Karsten Müller presents to you sensations from the world of endgames - partly reaching far beyond standard techniques and rules of thumb - and rounds off with some cases of with own examples.


Jonathan Speelman, born in 1956, studied mathematics but became a professional chess player in 1977. He was a member of the English Olympic team from 1980–2006 and three times British Champion. He played twice in Candidates Tournaments, reaching the semi-final in 1989. He twice seconded a World Championship challenger: Nigel Short and then Viswanathan Anand against Garry Kasparov in London 1993 and New York 1995.


Rules for reader comments


Not registered yet? Register