To win the Candidates tournament and become the challenger for the World Championship, Sergey Karjakin had to come back from defeat, or near defeat, time and again. It is a similar path that Alexander Khalifman took 17 years ago.

Sergey Karjakin will play Magnus Carlsen for the World Championship this November, and, if he wins, chess fans will remember his best games from that match and from the Candidates tournament –notably his victory in the last round over Fabiano Caruana. Those successes will go down in history, but there are some earlier games that were also important in his march to the championship.

To even reach the Candidates tournament, Karjakin had to reach the finals of the World Cup. (He did more than that, he won an unbelievable final match against Peter Svidler.) To get to the final, he had to cross a harrowing road. Twice – against Pavel Eljanov of Ukraine in the semi-finals and against Alexander Onischuk of the United States in the second round – he was on the verge of elimination. In both cases he lost the first game of a two-game mini-match and needed to win to stay alive.

Karjakin’s resilience reminded me of Alexander Khalifman’s performance in the 1999 FIDE World Championship tournament in Las Vegas. All the world’s strongest players, with the exception of Garry Kasparov (who held a competing world championship title) and Viswanathan Anand (who was rumored to be Kasparov’s challenger for the other title) participated.

Khalifman was a very strong Russian grandmaster who had not lived up to his considerable talent, and by that point he was starting to transition from being an active player to a trainer and writer. Despite being only the 32nd seed, he won the tournament, defeating Dibyendu Barua of India; Gata Kamsky of the United States; Karen Asrian of Armenia; Boris Gelfand of Israel; Judit Polgar of Hungary; Liviu-Dieter Nisipeanu of Romania; and Vladimir Akopian of Armenia.

The format was similar to that of the World Cup won by Karjakin, with the first five rounds consisting of two-game mini-matches (with rapid and blitz playoffs as needed). In each of the first two rounds, Khalifman lost the first game. Indeed, against Barua, Khalifman was in constant trouble, only getting through in six games. In all, Khalifman lost five games over the course of the event, but in every case but one, he came back to win the very next game. (In the sixth case, he drew Game 5 against Barua after the latter had tied the match in Game 4 to force a new set of tiebreak games. Khalifman then won Game 6.)

It was a remarkable performance by Khalifman, whose rating soon shot up to 2700 – which at that time was even more of a sign of elite status. Khalifman’s ability to fight back was exceptional, just as it is with Karjakin, and that ability has two benefits. First, there’s the concrete value on the score table in the form of the extra half and full points the player accrues. There’s also the benefit that comes from believing in one’s self: confident players are more likely to play better, which reinforces their confidence, which makes them play better, etc. In Khalifman’s case, there was likely a third benefit: the extra games helped him play himself into form and made him psychologically stronger when it was time to face the top seeds. (By contrast, there are the examples of Veselin Topalov in the championship played in Libya in 2004, or Eljanov in last year’s World Cup. Both started with incredible winning streaks, but lost their first tough match.)

Khalifman’s first game against Barua illustrates how out of shape he was, whereas several rounds later he had impressive victories over Polgar and Gelfand.

Barua, Dibyendu vs. Khalifman, Alexander
FIDE-Wch k.o. | Las Vegas | Round 1.1 | 31 Jul 1999 | ECO: B58 | 1-0
1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Nc6 6. Be2 e5 7. Nf3 h6 8. O-O Be7 9. Re1 O-O 10. h3 Be6 11. Bf1 Nb8
11... Rc8  )
11... Qa5  )
12. a4
12. b3  )
12... Nbd7 13. a5 a6 14. Nd5 Nxd5 15. exd5 Bf5 16. c4 Bg6 This variation has been around for ages, and in broadest outline the plans are for each player to use his pawn majority: White will play b4 and look to break with c4-c5, while Black will play ...f5 and use the e- and f-pawns to spearhead kingside play.
17. Nd2 A novelty at the time; White had hitherto played the immediate b4.
17... Bf6 I can't see anything wrong with the obvious
17... f5  )
18. Ra3 Re8 Clearly intending ...e4, so White prevents it.
19. Ne4
19. Bd3 is also good.  )
19... Bxe4?!
19... Be7 is probably better, intending ...f5, but it would require admitting 17...Bf6 was inaccurate.  )
20. Rxe4 Bg5 Clearing the path for the f-pawn.
21. Bxg5 hxg5 22. Rg4 The simple
22. b4 f5 23. Re1 is also good. White's queen will go to h5 and the rook to g3.  )
22... f5 23. Rgg3 g6 24. Qd2 Black's problem is that he really doesn't want to play ...f4, as that surrenders some important light squares. This is the result of the exchange on e4, which is in turn the result of playing ...Bf6 and not admitting the error.
24... f4 25. Rgb3! Rb8
25... Nc5 26. Qc2! Kg7 27. Rc3 White will recoup the tempo with b4.  )
26. Be2 Kg7 27. Bg4 Re7
27... e4  )
28. Bxd7! Rxd7 29. Rb6 Rc8 30. Qd3
30. Rc3!  )
30... Qf6 31. Rab3 Qf7 32. Rc3
32. f3  )
32... Qf5 33. Qd1 Qe4
33... Kh6! followed by ...e4 has Black right back in the game.  )
34. Qg4
34. b3!  )
34... Qf5! 35. Qd1 Qe4
35... Kh6!  )
36. b3! Rcc7 37. f3 Qf5 38. Qe1 White hasn't played perfectly, but he has finally achieved the coordination he needs. Black's kingside play has been terminated, while White retains the ability to execute his thematic plans with b4 and c5. Black's position is strategically lost.
38... Kh6 39. Qe4 Qxe4 40. fxe4 g4 The best try.
41. hxg4 Kg5 42. Kf2 Kxg4 43. c5! A nice tactical shot, taking advantage of the loose g-pawn and Black's surprisingly precarious king.
43... g5
43... Rxc5 44. Rxc5 dxc5 45. Rxg6+ Kh5 46. Re6  )
43... dxc5 44. Rxg6+ Kh5 45. Re6 Re7 46. Rh3+ Kg5 47. Reh6 Rg7 48. Kf3 Threatening mate in one.
48... Rg6 49. R3h5+ Kf6 50. Rxg6+ Kxg6 51. Rxe5 Black is completely lost.  )
44. c6 Rh7 45. b4! Rcf7 46. Rf3?
46. Rh3! is an easy win. Black cannot take twice, as White plays cxb7 and promotes before Black creates any threats. On the other hand, if Black doesn't take on h3, then he can't create any threats - he has no open lines and his king prevents the g-pawn from advancing.  )
46... Rc7?
46... bxc6 47. dxc6 d5! 48. exd5 e4 49. d6! exf3 50. c7! Rxc7 51. gxf3+ Kf5 52. dxc7 Rxc7 53. Rxa6 Rc2+ 54. Ke1 g4 55. fxg4+ Kxg4 It appears that Black has enough counterplay for a draw.  )
47. Rb3?
47. Rh3!  )
47... bxc6 48. dxc6 Rh1 49. Rc3
49. Rxa6 Rc1 50. b5 Rc2+ 51. Kg1 Rc1+  )
49... Rb1 50. Rc2 Rb3?
50... Kh4! Black needs counterplay, and running the g-pawn is the only way to get it.
51. Rxa6 g4 52. Rb6 g3+ 53. Ke2 f3+! 54. Kxf3 Rb3+ 55. Ke2 Rf7 is one possible variation, leading to a perpetual check.  )
51. b5! axb5 52. Rb7 Rc8 53. c7 f3 54. gxf3+
54. a6  )
54... Rxf3+ 55. Ke2 Rh3 56. a6 b4 57. Rb8
57. a7  )
57... Rh2+ 58. Kf1!!
58. Kd3 Rh3+!  )
58... Rh1+
58... Rxc2 59. Rxc8 b3 60. a7 b2 61. Rb8 Rxc7 62. a8=Q Rc1+ 63. Ke2 b1=Q 64. Rxb1 Rxb1 65. Qa3 wins, as Black cannot construct a fortress.  )
59. Kg2 b3 60. Rc3
Khalifman, Alexander vs. Polgar, Judit
FIDE-Wch k.o. | Las Vegas | Round 5.1 | 12 Aug 1999 | ECO: D39 | 1-0
1. Nf3 Nf6 2. c4 e6 3. d4 d5 4. Nc3 dxc4 5. e4 Bb4 6. Bg5 c5 7. Bxc4 cxd4 8. Nxd4 Bxc3+ 9. bxc3 Qa5 10. Bb5+ Bd7
10... Nbd7  )
11. Bxf6 gxf6 12. Qb3 Nowadays
12. Bxd7+ is the main move.
12... Nxd7 13. O-O a6 14. Rb1 Qc7 15. Qh5 Nc5 etc.  )
12... a6 The reason Khalifman's 12th move has gradually fallen out of favor is that
12... Bxb5 13. Nxb5 O-O 14. O-O Nc6 has been a drawing machine for Black, with
15. c4 Rad8 16. Qg3+ Kh8 17. Qh4 Kg7 18. Qg3+ Kh8 19. Qh4 Kg7 and a draw either now or after another repetition being a fairly popular way to conclude a short day at the office.  )
13. Be2 Nc6 14. O-O Qc7 15. Rab1 Na5 16. Qa3 Rc8 17. c4!? This was a near-novelty at the time. White had played it in the first game to reach the position, back in 1988, but in the intervening years White had always played 17.Rfd1 - even Kasparov preferred it. So this move may have come as a surprise to Polgar.
17... Qc5?! Black should face the danger and take the pawn.
17... Nxc4! 18. Bxc4 Qxc4 19. Rfd1 Qc3! leaves White sufficient compensation for the pawn, but not more. Objectively, should probably forget about "compensation" and go into an ending with equal material, either by taking on c3 and then b7 or like this:
20. Rb3 Qc5 21. Qxc5 Rxc5 22. Rxb7  )
18. Qc3 e5 19. Nb3 Nxb3 20. axb3 With a healthy pawn structure and a better king, the advantage is clearly White's.
20... a5 21. Rfd1 Be6 22. h3! A great move, aiming to trade bishops. Black's bishop on e6 is the glue holding her position together, so White aims to remove it.
22... O-O 23. Qg3+ Kh8 24. Qh4 Qe7 25. Bg4 Rg8 26. Rd3
26. Rd5!? Rxg4 27. hxg4 Bxd5 28. exd5 Kg7 29. Rd1  )
26... Rg5 27. Bxe6 Qxe6
27... fxe6 keeps the intruders out of d5, but weakens the second rank. White is nearly winning after
28. Rbd1 Rc7 29. Rd6  )
28. Rbd1 Rcg8 29. Rd6! Qc8
29... Rxg2+ 30. Kf1 Qe7 31. Rd7 Qf8 32. Rd8 Qe7 33. R1d7 Qe6 34. Rd6 Qe7 35. R8d7 Qf8 36. Qxf6+ Qg7 37. Rxf7 Qxf6 38. Rdxf6  )
30. g3 Kg7 31. R1d5 Setting up a nice trick, to which there seems to be no (sensible) defense.
31... h6
31... Qc7 32. c5 h6 33. Rxe5!  )
32. Rxe5! Rd8
32... fxe5?? 33. Qxh6#  )
32... Rxe5? 33. Qxf6+ Kf8 34. Qxe5  )
33. Red5!
33. Rxf6!! is a spectacular blow found by the computer. It's only slightly better than Khalifman's move, and as it involves much more risk with little reward (especially if White didn't have much time on his clock), it's right for a human to eschew it in a game of this magnitude.
33... Kxf6 34. f4  )
33... Rxd6 34. Rxd6 Rg6 35. Qf4 a4 36. bxa4 Qxc4 37. Rd8 Qc3 38. Kg2 Qb4 39. Ra8 h5 Rightly fishing for counterplay in a lost position, but this move expedites the end.
40. Qb8! Qxe4+ 41. Kh2 Rg5 42. Qh8+ Kg6 43. Rg8+ Kf5 44. Qh7+ A great win by Khalifman against an elite opponent.
44. Qh7+ Rg6 45. Qxh5+ Rg5 46. Rxg5+ fxg5 47. Qxf7+ Ke5 48. Qe7+  )
Khalifman, Alexander vs. Gelfand, Boris
FIDE-Wch k.o. g/25+10 | Las Vegas | Round 4.3 | 11 Aug 1999 | ECO: B92 | 1-0
1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be2 e5 7. Nb3 Be7 8. Be3 Be6 9. Nd5
9. O-O  )
9... Nbd7
9... Nxd5 10. exd5 Bf5 is another main line, with play roughly similar to what we saw in Barua-Khalifman.  )
10. Qd3 Bxd5 Gelfand has remained faithful to this approach.
10... O-O is an equally popular alternative.  )
11. exd5 O-O 12. g4!?
12. O-O Nc5 13. Qd2 Nfe4 14. Qb4 a5 15. Qc4 Rc8 16. Qb5 b6 17. f3 Nf6 18. Rfd1 Nh5 19. g3 g6 20. Nxc5 bxc5 21. Bd2 Ra8 22. f4 Ng7 23. fxe5 dxe5 24. Be3 Rb8 25. Qc4 Nf5 26. Bf2 Rxb2 27. Rab1 Rb4 28. Rxb4 cxb4 29. Qc6 Bd6 30. c4 bxc3 31. Qxc3 e4 32. Bg4 Bb4 33. Qb3 e3 34. Bxe3 Nxe3 35. Qxe3 Re8 36. Qd4 Bd6 37. Kg2 Qc7 38. Rd2 h5 39. Re2 Rxe2+ 40. Bxe2 Qc2 41. Qc4 Qd2 42. a4 Qe3 43. Qd3 Qe5 44. Bf3 Kg7 45. Bd1 h4 46. Qf3 hxg3 47. hxg3 Qd4 48. g4 Qd2+ 49. Kf1 Bc5 50. Bb3 Qc1+ 51. Ke2 Qb2+ 52. Kd1 Bb4 53. Qe3 Bd6 54. Bc2 Qa1+ 55. Ke2 Qh1 56. Qc3+ Kg8 57. Qc8+ Kg7 58. Qc3+ Kg8 59. Qc8+ Kg7 60. Qc3+ Kh7 61. Qf3 Qh2+ 62. Kd3 Kg8 63. Qe4 Bb4 64. Kc4 Qc7+ 65. Kb5 Qb7+ 66. Kc4 Qc8+ 67. Kd3 Qc3+ 68. Ke2 Qe1+ 69. Kf3 Qf1+ 70. Ke3 Bd6 71. Kd4 Qf2+ 72. Kd3 Bb4 73. Qe3 Qg2 74. Qe4 Qd2+ 75. Kc4 Qc3+ 76. Kb5 Qc5+ 77. Ka6 Qc8+ 78. Kb5 Bd6 79. Bd1 Qc5+ 80. Ka6 Qc8+ 81. Kb5 Kh7 82. Bc2 Qc5+ 83. Ka6 Bc7 84. Kb7 Qb6+ 85. Kc8 Bd6 86. Qc4 Qb8+ 87. Kd7 Qf8 88. Qe4 Qb8 89. Qc4 Qf8 90. Qe4 Qb8 91. Qc4 - (91) Anand,V (2773)-Gelfand,B (2761) Zuerich 2014 (rapid)  )
12... Nc5?!
12... Nb6 with the idea of ...Na4 has been played since then, and was suggested by Korchnoi at the time.  )
12... a5  )
12... e4!? was mentioned by Ftacnik in his notes to this game, weakening d4 but giving e5 to the knight.  )
13. Nxc5 dxc5 Without the pawn move to g4, White would play c4, castle short and go for queenside play. Here that option is less practical.
14. O-O-O e4 The stereotypical blockading idea
14... Ne8 , hoping to play ...Nd6, is too slow.
15. d6! Qxd6 16. Qf5 Qb6 17. h4  )
15. Qd2 Bd6 16. g5 Nd7 17. h4 Ne5 18. h5 Rc8?
18... f5! 19. gxf6 Qxf6 20. Rdg1 b5  )
19. Rh4! Winning an important pawn. Gelfand now goes full blast on the attack.
19... c4 20. Rxe4! Not fearing Black's next move.
20... c3 21. bxc3 Qa5 22. Kb1! Rxc3 23. Bd4 Rfc8 24. Bxc3 Rxc3 25. Qd4 f6 26. gxf6 gxf6 27. f4 Bc5 28. Qa4 Qb6+
28... Qxa4 is only "better" in a meaningless computer sense. It may take more moves for White to win from here, but with queens off the win is trivially easy for White.  )
29. Ka1 Nf7 30. Qe8+
30. Rg1+! Bxg1 31. Qe8+ Kg7 32. Re7 forces a speedy mate, but it's hard to find Rg1+ in a rapid game.  )
30... Kg7
30... Bf8 31. Re6  )
31. h6+!
31. Rg1+! works here too.  )
31... Nxh6 32. Qd7+
32. Rg1+! And here.  )
32... Nf7 33. Rb1
33. Rg1+!  )
33... Qd8 34. Rxb7 Qxd7 35. Rxd7 Rxc2 36. Bh5 Khalifman's human approach is plenty sufficient. Gelfand tries one last idea before giving up.
36... Ba3 37. Re1 Bb4 38. Rxf7+ Kh6 39. Rh1

Karjakin’s resilency may not be enough to beat Carlsen this fall, but he’ll definitely need it if he has any chance to succeed.


Dennis Monokroussos is a FIDE master who has written about chess on his blog “The Chess Mind,” since 2005. He has been teaching chess for almost 20 years and for the last 10 years has been making instructional chess videos, which can be found at Between 1995 and 2006, he taught philosophy, including a four-year stint at the University of Notre Dame.