Global Chess League day 6: two errors with the clock full, and two one-move punishments
Anand went wrong with nine minutes on his clock, Nepomniachtchi with eleven. In both cases the refutation was the only move that won, and in both cases it was found in under 35 seconds. We measured both games and recounted the day's four scores.
12 min read · Published on 2026-09-11
Short answer
Day six of the 2026 Global Chess League, on 10 September, left three teams tied on 15 match points with a single game point separating them. The day's two icon boards ended the same way: Viswanathan Anand handed over an equal position at move 28 with 9:13 on his clock, and Ian Nepomniachtchi handed over a playable one at move 29 with 11:38. This piece measures both games with Stockfish 19, shows that in each the punishment was the only move that won, and recounts the day's four scores from the boards themselves.
Event: Tech Mahindra Global Chess League 2026
How this was compiled: Compiled on 11 September 2026, with day six complete — rounds 21 to 24, played on 10 September. Clocks read move by move from the official broadcast PGN; evaluations measured by us with Stockfish 19 at depth 24 with three lines per position, always from White's point of view. The day's four match scores were recounted by us board by board using the league's scoring (3 points for a win with White, 4 with Black, 1 for a draw). The standings quoted are the ones FIDE published at the end of day six.
The essentials in a few lines
- At move 28 of Carlsen-Anand the position measured 0.00 at depth 24; after 28...Nd4 only 29.Rg6+ won — 29.Rxf8+ and 29.Qf2 both measure 0.00, and 29.Rg6+ measures +7.37.
- Anand played the losing move with 9 minutes 13 seconds on his clock against Carlsen's 5 minutes 43: this was not time trouble, it was a calculation error with the clock in his favour.
- Carlsen's 25.Bxh6 was obligatory rather than optional: it measures +0.38, while the engine's two alternatives, 25.Be3 and 25.Bb2, measure -1.01 and -1.15.
- Nepomniachtchi played 29...Ne4 in 12 seconds with 11:38 on the clock; the engine gave +0.69 to 29...g6 and +2.54 to its second choice, and after 29...Ne4 only 30.Rxd5 worked, at +4.57 against +0.60 for the alternative.
- Sindarov handed the game back once: 35.g5, played in 7 seconds, fell from +6.48 to +0.44, and he only won again because Nepomniachtchi erred a second time with 37...Rb7, when only 37...hxg5 held.
Day six, recounted board by board
What were the day 6 results at the 2026 Global Chess League?
Four matches, all on 10 September: Alpine APL Pipers 4-12 Ganges Grandmasters, Triveni Continental Kings 5-14 CheQ Mumba Masters, Alpine APL Pipers 11-6 PBG Alaskan Knights and FYERS American Gambits 13-5 Ganges Grandmasters. The day ended with Alpine, FYERS and Ganges tied on 15 match points, one game point apart.
The table at the end of the day is the tightest the league has produced so far: Alpine APL Pipers and FYERS American Gambits on 15 match points and 73 game points, Ganges Grandmasters on 15 and 72, and CheQ Mumba Masters on 12 match points and 73 game points — which means the side that has scored most on the boards sits fourth, because match points and game points measure different things. The Triveni Continental Kings are out of contention for the final.
We recounted the four scores from the six boards of each match, applying the league's scoring: 3 points for a win with White, 4 for a win with Black, 1 point each for a draw. All four matched the numbers FIDE published. Chess.com's bulletin gave 3-15 in Triveni against CheQ; the boards add up to 5-14, which is also FIDE's figure.
The detail that matters for the race to the final is that the Ganges Grandmasters played twice on the day, winning heavily in the morning and losing heavily in the afternoon — 12-4 over Alpine and 5-13 to FYERS. The icon-board games from those two matches are the subject of the sections below, and they have more in common than the scores suggest.
| Round | Match | Score | Icon board |
|---|---|---|---|
| 21 | Alpine APL Pipers - Ganges Grandmasters | 4 - 12 | Carlsen ½ - ½ Nepomniachtchi |
| 22 | Triveni Continental Kings - CheQ Mumba Masters | 5 - 14 | Firouzja 0 - 1 Vachier-Lagrave |
| 23 | Alpine APL Pipers - PBG Alaskan Knights | 11 - 6 | Carlsen 1 - 0 Anand |
| 24 | Ganges Grandmasters - FYERS American Gambits | 5 - 13 | Sindarov 1 - 0 Nepomniachtchi |
Carlsen and Anand: the sacrifice on h6 was compulsory
What was behind Carlsen's sacrifice against Anand on day six?
25.Bxh6 was not a choice between an attack and a quiet position: it was the only way for White not to stand worse. The engine measures the sacrifice at +0.38 and the two alternatives, 25.Be3 and 25.Bb2, at -1.01 and -1.15. Carlsen thought for 114 seconds and played it; Anand recaptured in 3.
The game was a King's Indian Attack (A07) with 1.g3 Nf6 2.Bg2 d5 3.Nf3 e6 4.O-O Be7 5.d3 O-O. The opening gave neither side an advantage, and the position had the classic shape of the variation: white pawns on d3 and e4, a good black piece on c5, play on both wings. Carlsen was the first to give ground, with 21.c4: after the move, Black's three best continuations all measure -0.69.
The move that opened the game came at move 23. With 23.h4 and 24.h5 Carlsen pushed the rook's pawn at the black king, and Anand answered 24...h6 after 34 seconds of thought. The move is human, and it is what almost every club player plays: stop the h-pawn before it reaches h6. The trouble is that it creates the target. The three continuations Stockfish 19 lists in the position are 24...Qd6 (-0.04), 24...f5 (0.00) and 24...Bd4 (0.00); 24...h6 does not make the list.
It is worth recording the right magnitude for the sacrifice, because it runs against the instinct of anyone reading the game backwards from the result. After 25.Bxh6 gxh6 26.Nxf6+ Rxf6 27.Rxf6 Rf8 the position is back to 0.00: White has regained the piece and has an attack that amounts to perpetual check if Black defends. 25.Bxh6 does not win the game; it stops Carlsen from standing a pawn's worth of evaluation behind. The win arrived three moves later, and not from this sacrifice.
| ♜ | ♜ | ♚ | |||||
| ♛ | ♟ | ||||||
| ♞ | ♟ | ♟ | |||||
| ♟ | ♝ | ♘ | ♟ | ♙ | |||
| ♟ | ♙ | ♙ | |||||
| ♙ | ♙ | ♙ | |||||
| ♔ | |||||||
| ♖ | ♗ | ♕ | ♖ |
Full line: 1.Bxh6 gxh6 2.Nxf6+ Rxf6 3.Rxf6 Rf8 4.Qf3
28...Nd4: the error that decided it, with nine minutes on the clock
What was Anand's decisive error against Carlsen?
28...Nd4, played after 44 seconds with 9 minutes 13 seconds on his clock. The position measured 0.00: 28...Qg7 equalised, 28...Be7 measured +0.19, and even 28...Rxf6 was playable at +1.18. After the move played, only 29.Rg6+ won — the engine's other two options, 29.Rxf8+ and 29.Qf2, measure 0.00, and 29.Rg6+ measures +7.37.
The diagram position is the test that day six set twice. On a simple count Black is a piece up — rook and two pawns against two minor pieces, with the white rook planted on f6 and the black king stripped of its pawn cover — and has to choose between defending and counter-attacking. 28...Qg7 is the defence: it offers the queen trade down the f-file, and after 29.Rxf8+ Bxf8 30.Qf2 Qf7 the endgame is level and riskless.
28...Nd4 is the counter-attack: the knight leaves with tempo against the f3-queen and threatens b3. What it misses is that the white queen does not have to move. 29.Rg6+ is check because the g7-pawn has not existed since the sacrifice, and after 29...Kh8 30.Rxh6+ Kg8 31.Rg6+ Kh8 32.Qe3 the white rook is untouchable on the sixth rank and the queen comes in on the diagonal. The detail that decides it is that the knight on d4 is too far away to come back.
The number that matters for anyone studying their own games is not the +7.37. It is the clock: 9 minutes 13 seconds, against Carlsen's 5 minutes 43. Anand did not err in time trouble; he erred with double his opponent's time and after 44 seconds of thought — the choice between defending and counter-attacking was made quickly because it looked easy, not because the clock forced it. Carlsen, at the other end, found the only move in 29 seconds. The game ended at move 37 with forced mate on the board.
| ♜ | ♚ | ||||||
| ♛ | |||||||
| ♞ | ♖ | ♟ | |||||
| ♟ | ♝ | ♟ | ♙ | ||||
| ♟ | ♙ | ♙ | |||||
| ♙ | ♙ | ♕ | ♙ | ||||
| ♔ | |||||||
| ♖ |
Full line: 1...Nd4 2.Rg6+ Kh8 3.Rxh6+ Kg8 4.Rg6+ Kh8 5.Qe3
Sindarov and Nepomniachtchi: twelve seconds to give up d5
How did Sindarov beat Nepomniachtchi at the Global Chess League?
With White, in a Queen's Gambit Declined (D37) where White's edge reached +1.43 at move 17 and had already melted to +0.48 by move 21. The error came at move 29: Nepomniachtchi played 29...Ne4 in 12 seconds, with 11 minutes 38 on his clock, and 30.Rxd5 — the only move that worked, at +4.57 against +0.60 for the alternative — settled the game.
The opening was the Exchange Variation of the Queen's Gambit Declined with 6.Bf4, and the first meaningful decision was Black's: 7...Nh5 8.h3 Nxf4 9.exf4, which surrenders the bishop pair and gives White an unusual pawn structure with pawns on f2 and f4. Sindarov played 12.Kf1 and 15.Kg2 to walk the king off the e-file without losing the rook's future, a rapid-chess device that only works because the centre is locked.
White's advantage appeared at move 17 with 17.Bf5, and the engine measured it at +1.43 after 17...Rc8. It melted again straight away: after 20.Re2 the position is back to +0.48, and the game settled into a rook-and-knight middlegame with a strong white knight on e6 and a weak black pawn on d5. It is exactly the kind of position where each individual move is worth little, until one is worth everything.
At move 29, after 29.f4, Black needed to free the king: 29...g6 measures +0.69 and is the only way to answer the pressure on f7 and h6 at once. The engine's next two options, 29...Ra8 and 29...Rb8, already measure +2.54 and +2.69. Nepomniachtchi played 29...Ne4 in 12 seconds with 11:38 on the clock: the knight looks active and hits the d3-rook, but it leaves d5 undefended. 30.Rxd5 is the only move that exploits it — 30.g5 and 30.h4 measure +0.60 and +0.58 — and Sindarov found it in 34 seconds.
| ♜ | |||||||
| ♜ | ♚ | ♟ | |||||
| ♟ | ♘ | ♞ | ♟ | ||||
| ♟ | ♟ | ♙ | |||||
| ♙ | ♙ | ||||||
| ♙ | ♖ | ♙ | |||||
| ♖ | ♔ | ||||||
Full line: 1...Ne4 2.Rxd5 Nc5 3.Rc2 Rb8 4.Nxc5
The move where Sindarov gave it all back
Did Sindarov also go wrong against Nepomniachtchi?
He did, decisively: 35.g5, played in 7 seconds, dropped the evaluation from +6.48 to +0.44. The engine wanted 35.Rd7+ or 35.Kh4, both at +6.48. The game only became a white win again because Nepomniachtchi erred a second time two moves later with 37...Rb7, when only 37...hxg5 held.
The endgame is rooks with pawns on both wings, the kind of position where the evaluation moves faster than the board appears to. After 34...Rb2 White has two active rooks, an extra pawn and an advanced king. 35.Rd7+ drives the black king off the seventh rank before anything else; 35.Kh4 prepares g5 with the king supporting it. Both measure +6.48.
35.g5 pushes the pawn without preparing anything, and the reply 35...R2xb3 trades a pair of rooks at the precise moment White needed to keep both. After 36.Rxb3 Rxb3+ 37.Kg4 the position measures +0.44 — near equality, in an endgame where the player with more time was Nepomniachtchi, on 5:51 against 4:47. Sindarov spent 7 seconds on the move that threw away six points of evaluation.
The next error was Black's, and it was the last. With 37...Rb7, played after 84 seconds, Nepomniachtchi missed the only continuation that saved, 37...hxg5, which holds +0.44; 37...g6 and 37...Rb8 measure +7.06, and the move played is not among the top three. The game ended at move 45. It is the same lesson as the other two: the serious error arrived with clock in hand, and the punishment was unique.
| ♜ | |||||||
| ♚ | ♟ | ||||||
| ♟ | |||||||
| ♟ | ♟ | ♖ | ♙ | ||||
| ♙ | ♙ | ||||||
| ♙ | ♖ | ♔ | ♙ | ||||
| ♜ | |||||||
Full line: 1.g5 R2xb3 2.Rxb3 Rxb3+ 3.Kg4 Rb7 4.Rxc5
What day six shows about errors and the clock
Does time trouble explain the day six errors at the Global Chess League?
It explains none of the day's three decisive errors. Anand had 9:13 when he played 28...Nd4, Nepomniachtchi had 11:38 when he played 29...Ne4, and Sindarov had 4:48 when he played 35.g5 — all of them above a minute per remaining move. What the three share is not the clock: it is choosing the active move in a position that asked for the solid one.
The three errors are of one type, and it is a type that shows up in club games just as often. In each, the move played sends a piece to an advanced square — knight to d4, knight to e4, pawn to g5 — and each is the first move that occurs to a player who is happy with their own position. None of the three needed long calculation to be discarded: one round of asking what the opponent replies would have done it.
The counterpoint is how quickly the punishments were found. 29.Rg6+ took 29 seconds and 30.Rxd5 took 34, and in both cases they were the only moves that won. That is the asymmetry that defines rapid chess at this level: an error costs dearly because the other side does not need much time to find the refutation — it needs to be looking for one.
For anyone studying their own games, day six suggests a cheap and simple filter: look for the moves where you had time to spare and played quickly anyway. That is where the error your engine analysis marks as incomprehensible tends to live — and, as day six shows, it is the same address for players rated 2721 and 2729.
Frequently asked questions
Who leads the 2026 Global Chess League after day six?
Three teams tied on 15 match points: Alpine APL Pipers and FYERS American Gambits, both on 73 game points, and Ganges Grandmasters on 72. CheQ Mumba Masters have 12 match points and 73 game points, and the Triveni Continental Kings have lost their chance of reaching the final.
Did Carlsen beat Anand on day six of the Global Chess League?
Yes, with White, in round 23, in a 37-move King's Indian Attack that ended with forced mate on the board. The game was level until move 28, when 28...Nd4 turned a 0.00 position into +7.37, and the refutation, 29.Rg6+, was the only move that won.
Why was Carlsen's 25.Bxh6 not an attacking sacrifice?
Because it does not produce an advantage: it measures +0.38, and the position is back to 0.00 four moves later with the piece regained. It was the best move by a wide margin — the engine's alternatives, 25.Be3 and 25.Bb2, measure -1.01 and -1.15 — but in the job of holding the balance, not of winning the game.
How were the evaluations in this article measured?
With Stockfish 19 running locally, at depth 24 with three lines per position, always from White's point of view; the broadcast's automatic annotation was used only to locate the moves worth measuring. The clocks come from the broadcast PGN, and the match scores were recounted by us from the six boards of each round.
Update log
- 2026-09-11 — First publication, with both icon-board games measured by us at depth 24 and the day's four match scores recounted from the boards.
Sources
- GCL 2026, Day 6: One game point separates Alpine APL Pipers, FYERS American Gambits and Ganges Grandmasters at the top — FIDE
- 2026 Global Chess League Day 6: Alpine Pipers, American Gambits, Ganges Grandmasters Share 3-Way Lead — Chess.com
- Alpine APL Pipers - PBG Alaskan Knights, round 23, with move-by-move clocks — Lichess
- Ganges Grandmasters - FYERS American Gambits, round 24, with move-by-move clocks — Lichess
Related guides
- King's Indian Attack vs the French Defense and the Caro-Kann: the setup, the plan, and what the engine says
- How to calculate candidate moves in chess: checks, captures and threats before you commit
- Recurring chess mistakes: how to find the pattern that repeats across your games
- How to avoid time trouble in chess: five rules for the last three minutes
Find your fast moves in positions that deserved time
The three errors of day six share one signature: an active move, played quickly, with clock to spare. Import your games and see which moves you decided fast in a position that was worth a minute.
Create a free account