How Carlsen saved a lost position against Nepomniachtchi with five seconds
Twenty moves lost, the clock under fifteen seconds and exactly one move that drew. The day-two game at the Global Chess League is a manual of practical defence.
11 min read · Published on 2026-09-07
Short answer
Ian Nepomniachtchi held almost six pawns of advantage against Magnus Carlsen on day two of the Global Chess League and the game ended in a draw after 89 moves. Carlsen played 52 of his 89 moves with under fifteen seconds on the clock and found, with five seconds left, the only move that drew. This article shows the two positions where the game turned, explains why active defence produces more chances than passive defence, and turns that into training you can apply.
Event: Tech Mahindra Global Chess League 2026
How this was compiled: Game from day two, played on 6 September 2026. Evaluations measured with Stockfish 19 at depth 24.
The essentials in a few lines
- White reached an evaluation of +5.96; the game ended in a draw after 89 moves.
- On move 30, 30.Qxe6 kept the decisive advantage; the capture chosen, 30.Rxe6, gave back more than three pawns of evaluation.
- On move 60, 60.Rd4 still won with +4.45; after 60.Kxe4, only 60...Re7+ drew — 60...Rf7 would have lost, at +3.09.
- Carlsen played 52 of his 89 moves with under fifteen seconds on the clock, and from move 38 onwards never had more than that.
- The draw did not decide the match: Alpine APL Pipers beat Ganges Grandmasters 11-6 and would have won even had they lost the top board.
How Carlsen got lost
What went wrong for Carlsen against Nepomniachtchi?
A Rossolimo Sicilian in which Black spent too long on a knight manoeuvre and ended up with an exposed king. From move 22 the evaluation never returned to equality, and by move 26 Black was objectively lost.
The opening was 1.e4 c5 2.Nf3 Nc6 3.Bb5 g6 4.O-O Bg7 5.Re1, and the first sign came immediately: 5...Nh6, a move that takes the evaluation from +0.24 to +0.76. Black followed with f5 and e6, a structure that gains space but leaves permanent holes on the light squares around the king.
Nepomniachtchi needed nothing special to accumulate the advantage: he traded the minor pieces defending the black king, put queen and rook on the open file and pushed d5 at the right moment. On move 26, with 26...Rfd8, the evaluation went from +1.76 to +4.26 — from there on Black was lost, not merely worse.
It is worth being honest about the nature of the game: this is not a defensive masterpiece against perfect play. It is a rapid game in which the winning side had several chances and missed some, while the losing side kept setting problems until one of them cost dearly. That is precisely why it teaches more than a clean win.
The first missed chance: which piece captures
Where did Nepomniachtchi first let the win slip?
On move 30. After 29...Qc5, the engine gives +5.96 for 30.Qxe6, capturing with the queen. Nepomniachtchi captured with the rook — 30.Rxe6 — and the evaluation fell to +2.77.
The two captures look equivalent and are not. With 30.Qxe6 the queen enters with threats and the rook on b6 keeps cutting the seventh rank; Black is left without counterplay and with a trapped king. With 30.Rxe6 the queen stays on b3, far from the black king, and the rook on e6 is exposed to tempo gains — and gaining tempi is exactly how a defensive position survives.
The difference between the two lines is not deep calculation: it is choosing which piece attacks and which stays on defensive duty. In a 20-minute game, with the opponent creating a threat every move, that is the kind of decision taken by habit rather than by variation — and the lazy habit is to capture with the lesser piece.
After 30.Rxe6 Kg7 31.Rxa6 Nxe5, Black had recovered a pawn and, more importantly, taken the knight out of passive defence. The evaluation stayed winning for White for many moves yet, but the nature of the position had changed: it was no longer an execution, it was a conversion.
| ♜ | ♚ | ||||||
| ♞ | ♟ | ||||||
| ♟ | ♖ | ♟ | ♟ | ||||
| ♛ | ♙ | ♟ | |||||
| ♕ | ♘ | ♙ | |||||
| ♙ | ♙ | ♙ | |||||
| ♔ |
Full line: 1.Rxe6 Kg7 2.Rxa6 Nxe5 3.Ra1 Kh6
Active defence: why Carlsen never sat still
What does it mean to defend a worse position actively?
It means choosing, between two bad moves, the one that forces the opponent to solve something. Passive defence minimises immediate risk and hands over the time; active defence accepts a slightly worse evaluation in exchange for keeping the opponent calculating.
The game shows a clear stretch of this between moves 31 and 55. A piece down, Carlsen put the king on h6 and then g7, sought exchanges of heavy pieces, pushed the kingside pawns and returned material to reach a rook endgame. None of those moves improves the evaluation — several make it worse — but every one of them demands a precise answer.
The effect shows on the clock, not in the evaluation. Nepomniachtchi finished move 60 with 33 seconds; Carlsen had 5. The player who was winning had to spend time on each of those decisions, and the 2-second increment from move 41 did the rest: in a position where every move requires checking, two seconds do not cover the bill.
It is important not to turn this into a magic recipe. Active defence does not save lost positions against precise play — it raises the chance of error in positions that demand precision for many moves. Here it demanded around thirty moves of precision, and White erred twice.
- Prefer the move that creates a threat, even if the evaluation gets slightly worse.
- Trade the pieces attacking your king; keep the ones that give counterplay.
- Returning material to reach a defensible endgame is a technical decision, not surrender.
- Count how many precise moves the opponent still needs: that is your real chance.
Move 60: one check, five seconds
Which move saved the draw?
60...Re7+. After 60.Kxe4, it was the only move that drew: everything else lost. The natural alternative 60...Rf7 leaves the evaluation at +3.09 for White. Carlsen had five seconds on the clock.
The position in the diagram is rook and two pawns against rook — usually winning, and the engine agrees: 60.Rd4 kept +4.45. But 60.Kxe4 allows the series of side checks that drives the white king away from the front of the pawns, and without the king escorting them, two flank pawns do not promote against an active rook and king.
What makes 60...Re7+ work and 60...Rf7 fail is the order. With the immediate check, the white king has to decide where to go before organising the pawns, and every square it chooses opens an entry for the black king. Without the check, White gains exactly the time needed to place rook and king on the right squares — and then there is no defence left.
The rest was technique: 60...Re7+ 61.Kd4 Kf3, the black king walks in, captures f2 on move 82, and the rook finally gives itself for the g7-pawn on move 89. A draw by insufficient material, with the last stretch of the game played, on the black side, almost entirely on the two-second increment.
| ♜ | |||||||
| ♟ | ♚ | ||||||
| ♔ | ♙ | ||||||
| ♖ | ♙ | ||||||
Full line: 1.Kxe4 Re7+ 2.Kd4 Kf3 3.Kc5 Re8 4.Kc6 Re1
What you can train from here
How do you train defence in worse positions?
With two exercises that need no engine: reviewing your own losses to find the move where you stopped creating problems, and drilling rook endgames with flank pawns until defending by side checks is automatic.
The first exercise is review. In a lost game, mark the move from which all your moves became answers to the opponent's threats. There is almost always an earlier point where an awkward alternative was available, and it was almost always discarded for making the evaluation worse. That is the decision to train, and it comes up far more often than a combination.
The second is pure technique. Rook and two flank pawns against rook is one of the most common endgames in rapid chess and one of the least studied by players between 1200 and 1800. Knowing that the defence starts by driving the enemy king away with checks — and why the order of the checks matters — is worth more points per hour of study than almost any opening line.
There is also the habit of counting chances instead of counting pieces. When you are worse, the useful question is not how much you are losing by, but how many precise moves the opponent still has to find. If the answer is thirty, as in this game, the position is far from decided — even with the clock against you.
Frequently asked questions
How did Carlsen draw a lost position against Nepomniachtchi?
By defending actively for around thirty moves and taking advantage of two mistakes. On move 30 White captured with the rook instead of the queen, giving back more than three pawns of evaluation (from +5.96 to +2.77); on move 60, 60.Kxe4 allowed 60...Re7+, the series of side checks that draws the rook endgame. The game ended level after 89 moves.
How do you defend a worse position without playing passively?
By choosing, among bad moves, the one that forces the opponent to solve a concrete problem. That includes returning material to reach a defensible endgame, trading the pieces attacking your king and keeping the ones that give counterplay. Passive defence lowers the immediate risk but hands the opponent the time needed to convert.
Is rook and two pawns against rook always a win?
No. With flank pawns and an active defending king, defence by side checks often holds: without the king escorting them, the pawns do not promote. In this game 60.Rd4 was still winning with +4.45, but 60.Kxe4 allowed the checks and the draw came in under thirty moves.
How much time did Carlsen have on the clock?
He played 52 of his 89 moves with under fifteen seconds, and from move 38 onwards never had more than that. At move 60, the decisive moment, he had five seconds against Nepomniachtchi's 33. The league's time control gives 20 minutes and a 2-second increment only from move 41.
Update log
- 2026-09-07 — First publication, with both critical positions checked against the engine.
Sources
- Ganges Grandmasters - Alpine APL Pipers: round five games — Lichess
- Gambits cling to top after a wild Day 2 in Bengaluru — Global Chess League
Related guides
Find out how you play when you are worse
Defence is a measurable habit: your games show the move from which you stop creating threats. Import them and compare what happens in your worse positions.
Create a free account