Miracles in Game Three

Those of you who have followed closely the matches from the World Cup remember how easily Svidler went through the semi-final. After a win with the black pieces in game one against Anish Giri, he simply did not let a single chance to his opponent in game two by choosing an ultra-solid opening line as White. Soon many of the pieces disappeared and the Russian GM comfortably went into the final.
Many people expected a similar scenario in game three of the final match. Svidler needed only a draw to win the event outright after the furious 2-0 start. At the beginning everything seemed fine with his strategy, although Karjakin tried to complicate matters at an early stage of the game:

[Event "FIDE World Chess Cup"]
[Site "?"]
[Date "2015.10.03"]
[Round "57.1"]
[White "Svidler, Peter"]
[Black "Karjakin, Sergey"]
[Result "0-1"]
[WhiteElo "2727"]
[BlackElo "2753"]
[Annotator "Bojkov, Dejan"]
[PlyCount "60"]
[EventDate "2015.??.??"]
[WhiteClock "0:12:32"]
[BlackClock "0:16:45"]

1. e4 c5 {In a must-win situation Karjakin probably did not even consider the
Berlin as an option.} 2. Nf3 d6 3. d4 cxd4 4. Qxd4 {Diagram [#] The last time
Svidler played this solid was back in 1999 against Garry Kasparov.} a6 5. c4
Nc6 6. Qe3 Nf6 7. h3 g6 8. Nc3 Bg7 9. Be2 Nd7 {[%cal Gg7a1] Diagram [#] Sooner
or later this knight will step back to open up the bishop and get to a better
position. Black can also start with the castling:} ({Relevant:} 9... O-O 10.
O-O Rb8 11. Rd1 Nd7 12. Rb1 Qb6 13. Qd2 Nde5 {as in Lagarde,M (2540)-Edouard,R
(2634) Saint Quentin 2015}) 10. Rb1 {White carefully finishes the development.}
Nde5 11. O-O O-O 12. Rd1 Nxf3+ 13. Bxf3 {Both sides achieved what they wanted-
approximate equality. However, we should not forget that Karjakin is in a
must-win situation.} f5 {The typical Accelerated Dragon counter-play.} 14. exf5
$146 {[%csl Re7][%cal Re1e7,Rd1d8] Diagram [#] And a standard reply which
appears to be a novelty. Svidler does not want to allow the f5-f4 advance. In
correspondence game another player with the white pieces did:} ({Predecessor:}
14. Nd5 e6 15. Nf4 e5 16. Ne2 f4 17. Qa3 Qg5 18. Rxd6 Bxh3 {with complications
in Szymanski,R (2372)-Chomicki,H (2229) ICCF email 2011. Karjakin definitely
would not mind these.}) 14... Bxf5 15. Be4 Qd7 16. Nd5 Qe6 17. Bxf5 Qxf5 {
Diagram [#] One more light piece disappeared. At least Black got some activity
along the half-open f file.} 18. Bd2 {Of course not} (18. Nxe7+ $4 Nxe7 19.
Qxe7 Rae8 20. Qxd6 Qxf2+ 21. Kh1 Re1+ 22. Rxe1 Qxe1+ 23. Kh2 Be5+ $19) 18...
Rae8 19. Bc3 e6 20. Nb6 d5 $1 {[%csl Yb6,Rg1] Diagram [#] The only chance to
complicate matters. Black starts an attack through the center while the white
knight is away from the main battlefield.} 21. Bxg7 Kxg7 {Third light piece
leaves the board...} 22. Qc5 ({Black is somewhat better in the endgame after}
22. cxd5 $6 exd5 23. Qc5 d4 24. Qxf5 Rxf5 25. Rd2 $15) 22... Rf6 {A risky
decision, the only real chance to play for a win! If this was an ordinary game
Karjakin would have opted for} (22... dxc4 23. Qxf5 exf5 24. Nxc4 Re7 $11 {
Diagram [#]}) 23. b4 {With the idea b4-b5. Once again the d5 pawn is poisoned
due to} (23. cxd5 exd5 24. Nxd5 Re2 25. Qb6 Qxf2+ 26. Qxf2 Rfxf2 {and the
rooks rock.}) 23... Ne5 24. cxd5 Nd3 {This was the position that Karjakin had
in mind. He will regain the pawn on f2 and open up the white king. True, in
the process his knight might get trapped, but did he have anything to lose?}
25. Qe3 ({Another way to defend is} 25. Qc7+ $5 {when Black has to choose
wisely where to go with the king. The obvious square} Kg8 $2 {is a bad choice
because of} (25... Kh8 {is better but then the line} 26. Qc3 Nxf2 ({White has
risk-free advantage in the line} 26... Qxf2+ 27. Kh1 e5 28. Qxd3 Qxb6 29. d6
$14) 27. Nd7 {is good for White.}) ({According to the computer, the best move
is} 25... Kh6 {but then White can win a tempo with a timely check against the
enemy king as in the line} 26. Qg3 Nf4 27. Qh4+ $1 {(away from the fork on e2)}
Kg7 28. dxe6 g5 29. Qg4 $11) 26. Nd7 Qxf2+ 27. Kh2 Rf7 28. Rf1 Qxf1 29. Rxf1
Rxf1 30. dxe6) 25... Nxf2 {The match situation is not about rational decisions
any more! Once again, from objective perspective best was} (25... Qxf2+ 26.
Qxf2 Nxf2 27. Re1 Nd3 28. Re3 Nf4 {Diagram [#] with a draw.}) 26. Rf1 {The
knight is trapped.} Qe4 27. Rbe1 {Svidler plays according to the position. He
had probably calculated the line which happened later in the game. A draw
could have been achieved by simple means with} (27. Qxe4 Nxe4 28. dxe6 Rfxe6
29. Nd5 Nd2 30. Nc7 $11) 27... exd5 (27... Ref8 28. Qxe4 Nxe4 29. Rxf6 Nxf6 $16
{leaves Black no hint of a chance.}) 28. Rxf2 $6 {Diagram [#] Not yet a
mistake but a huge step forward it. White could have won the knight with} (28.
Qd2 Nxh3+ 29. gxh3 Rxf1+ 30. Rxf1 {Diagram [#] when Black has chances for a
draw. But only in this game, not in the match...}) 28... Qh4 {Probably Svidler
was hoping for the nice line:} (28... Qxe3 29. Rxe3 Rxe3 30. Rxf6 Kxf6 31.
Nxd5+) 29. Qd2 $4 {[%csl Re1] Diagram [#] Unbelievable! The exhausting matches
and the final pressure came too much for Svidler. He blunders the game away.} (
{Instead} 29. Qxe8 Qxf2+ 30. Kh2 Qxb6 {would give some winning chances for
Black but should be still a draw with careful play. For example} 31. Re7+ Kh6
32. Rd7 Qxb4 33. Qg8 Qf4+ 34. Kh1 Qe5 35. Rxh7+ Kg5 36. Qd8 {followed by
Rh7-d7. Black cannot win with his king that exposed.}) 29... Rxf2 30. Qc3+ (30.
Qxf2 Rxe1+) 30... d4 {Diagram [#] Svidler resigned due to} (30... d4 31. Qc7+
Rf7) 0-1

Report.