Polina Shuvalova at the Global Chess League: five wins over grandmasters, and how many came from worse positions

Short answer

Polina Shuvalova, an international master rated 2398, won five of her six games over the first five days of the Global Chess League 2026, all against higher-rated grandmasters, and drew the other two of the seven played up to 10 September. This piece checks the run against the PGN and the engine: her White repertoire was 1.d4 in three of four games, with exchange variations, not the Italian that has been quoted; in three wins she was never worse; and both wins over Zhu Jiner came from worse positions — one of them lost by almost seven pawns after 44.Ra6, until Zhu handed it back on the next move.

Event: Tech Mahindra Global Chess League 2026

How this was compiled: Seven Global Chess League 2026 games, 5 to 10 September, read from the official broadcast PGN. Every position went through Stockfish 19 with a limit of 3 seconds each, at depths of 11 to 16; the decisive positions were measured again at depth 24, and those are the figures quoted. Evaluations are always from White's point of view.

The essentials in a few lines

• In the six games of the first five days, Shuvalova (IM, 2398) won five and drew one, and all five wins came against grandmasters rated 17 to 67 points higher.
• With White, Shuvalova played 1.d4 in three of four games and 1.e4 once, in the Exchange Variation of the Ruy Lopez; there was no Italian Game.
• In three of the wins she was never worse: her worst moment against Divya Deshmukh, Bibisara Assaubayeva and Kateryna Lagno was +0.10, −0.02 and 0.00.
• Against Zhu Jiner in round two there were three serious mistakes in three moves with both players under 1min15: 43...Bd5 (+5.1), 44.Ra6 (−6.8) and 44...Nc4 (+3.2), at depth 24.
• In round eighteen, Zhu Jiner had a perpetual check with 37.Ng4 and played 37.Nf4, with 58 seconds, in a position that went from 0.00 to −8.3.

The run: five wins with the lowest rating on her board

Shuvalova plays on board 4 for Ganges Grandmasters, one of the two boards reserved for women, and she is the lowest-rated of the six players on board 4 across the six teams. That makes the result more remarkable than the tally suggests: in all seven games her opponent was a grandmaster rated between 17 and 67 points higher.

Under the league's scoring, the five wins produced 16 game points — four with White, at 3 points each, and one with Black, at 4 — and the two draws added 2 more. That did not always decide the match: in round five she beat Divya Deshmukh and Ganges lost 11-6 to Alpine APL Pipers, and in round ten she drew with Harika Dronavalli in a match the team lost 12-3.

The table below lists all seven games, with the opening as classified by the broadcast and the match score. The evaluations quoted in the following sections were measured by us; the broadcast PGN also carries automatic evaluations, which were used only to locate the moves to measure.

The repertoire: 1.d4 and exchange variations, not the Italian

The reading of her run that has been quoted — clear plans in the Italian and the Ruy Lopez with 1.e4 — does not match the PGN. Her only 1.e4 game with White was the Ruy Lopez with 4.Bxc6, the Exchange Variation, against Kateryna Lagno. The other three began with 1.d4, and two of them were Queen's Gambits Declined: the Exchange Variation with 4.cxd5 against Zhu Jiner and the Cambridge Springs against Bibisara Assaubayeva.

What survives of the thesis is something else, and more useful: her choices with White reduce theory on purpose. The Exchange Variations of the Queen's Gambit and of the Ruy Lopez fix the pawn structure in the first few moves and move the game into familiar plans — the minority attack in one case, the pawn majority in the other — rather than forcing lines that have to be memorised. Against the Benko she accepted the pawn and chose the Modern Variation with 5.e3, instead of the lines in which White tries to hold on to everything at any cost.

For anyone building a club repertoire, the lesson is not to copy the openings but the criterion: choose variations whose structure you understand and whose plans repeat from game to game. This site's guides to a 1.e4 e5 repertoire and to the English Opening apply the same reasoning to other systems.

Three wins without ever being worse

The method has two stages, and they are worth explaining because the figures depend on them. First, every position from the seven games went through Stockfish 19 with a limit of 3 seconds each, which gives depths between 11 and 16 — enough to say whether someone was worse, not to publish the exact evaluation of a move. Then the decisive positions were measured again at depth 24, and those are the figures quoted as move evaluations in this piece. All evaluations are from White's point of view: a positive number favours whoever has White, whatever Shuvalova's colour.

The three games share the same shape. Against Divya Deshmukh, in the Benko, Shuvalova had a small edge, between +0.1 and +0.7, from move 10 to 20, and it grew to +1.5 before 25...Rb1, played by Divya with 88 seconds, which took it to +5.0 at depth 24 — the move that kept the game going was 25...Bd5. Against Bibisara Assaubayeva, in the Cambridge Springs, the evaluation stayed between 0.00 and +1.4 until 23...Nxd5, which the measurement takes from +1.3 to +4.1, when 23...Rxc1 held — and that mistake was not about the clock: Assaubayeva still had 5 minutes 21 seconds. Against Kateryna Lagno, in the Exchange Variation of the Ruy Lopez, the evaluation stayed between 0.00 and +0.9 from move 10 to 30, and the knight endgame slipped away without a single mistake: at depth 24 the position was already worth +4.7 before 37...Nxd5+, played with 44 seconds, which took it to +6.1. The quick scan showed only +1.6 at the same point — in an endgame with knights and pawns, depth changes the reading. The f-pawn queened nine moves later.

What the three show is low-risk patience, not tactical brilliance. None of the positions required a difficult move from Shuvalova to stay level or better, and in two of them the opponent's decisive moves came with less than a minute and a half on the clock. The contrast is her draw with Harika Dronavalli in round ten: there it was Shuvalova who was worse, with White reaching +1.85 after 23...Rb8, and she held the position for almost forty more moves.

Zhu Jiner, round two: an advantage lost in one move and handed back the next

The game came out of an Exchange Variation of the Queen's Gambit with 8.Qf3 and simplified until 33...Rxc2 34.Rxc2, when rook faced bishop and knight, with seven white pawns against five black. From move 25 to 39 the scan gives 0.00 in practically every position. What changed in that stretch was the clock: Zhu reached move 36 with 95 seconds against Shuvalova's 188, and the league's time control gives no increment before move 41.

White's advantage was born from Zhu's moves with a short clock — 39...Ke6, with 83 seconds, and 40...a4, which opens the queenside — and swung from there. On move 43, with 53 seconds, Zhu played 43...Bd5, and the measurement at depth 24 gives +5.1, against +0.7 for 43...b5. Shuvalova, with 73 seconds, replied 44.Ra6, which loses: after 44...Nc8!, the bishop goes to b7 and the rook has no square, and the evaluation goes to −6.8. The move that kept everything was 44.Rh7, at +6.2. Zhu missed the resource and played 44...Nc4, which handed back +3.2: three serious mistakes in three moves, with both players under a minute and fifteen seconds.

From move 46 to the end both played with less than 40 seconds and a 2-second increment per move. The scan shows large swings in that stretch, which we do not publish move by move: at such shallow depth in an endgame with minor pieces, the number swings because of the measurement's limits, not only because of the players' mistakes. What holds is the result. Shuvalova converted on move 70, and the win was worth the match: with a draw instead, Triveni would have won 10-8.

Zhu Jiner, round eighteen: worse for twenty moves, and castling on move 34

The opening gave White what the Trompowsky usually promises: after 1.d4 Nf6 2.Bg5 e6 3.e4 h6 4.Bxf6 Qxf6 5.Nc3 Nc6 6.Nf3 Bb4 7.Qd3 d5 8.e5 Qd8, White gained space with the pawn on e5, and the black king stayed on e8. Zhu pushed the h-pawn to h5 and kept up the pressure for fifteen moves. The scan gives White more than a pawn's advantage for much of that stretch.

The most dangerous moment came with 21...Bc6 22.Rh3 Qd7 23.Rg3 Bf8: the white rook reached the g-file, eyeing g7, and the black bishop retreated to defend. At depth 24, the position after 23...Bf8 is worth +2.7 for White; 23...b4, opening the queenside at once, held at +2.2. On the next move, instead of reinforcing the attack, Zhu retreated the knight with 24.Ne2, and the evaluation fell to +0.6 — the engine wanted 24.Nh4 Rb6 25.Bg4, with a sacrifice on g6 in the air. Black opened the queenside with ...b4 and ...b3, and only on move 34, after 33...Ba3 34.bxa3, did Shuvalova castle — with check, 34...O-O+.

The ending was decided in three moves. With both clocks under a minute and a half, 35.Rf3 Rxf3+ 36.gxf3 Nxa3 left the position at 0.00, and the engine shows Zhu had a perpetual check: 37.Ng4 b2 38.Nxh6+ Kh7 39.Bc2+ Nxc2 40.Qxc2+ Kh8 41.Nf7+ Kg8 42.Nh6+. With 58 seconds, Zhu played 37.Nf4, and the evaluation went to −8.3: the b3-pawn runs through to queen. Shuvalova's previous move, 36...Nxa3, was not the best either — 36...Qxa3 held −1.4 — and it was what gave Zhu the chance of a draw. The black queen came in via a2, Zhu played 38.Ke2 with 19 seconds, and the game ended on move 47. The win with Black brought 4 game points to Ganges' 15-3, which equalled the widest score of the tournament up to day five.

Frequently asked questions

Update log

• 2026-09-10 — First publication, with all seven games read from the broadcast PGN and evaluations measured by us.

Sources

• Tech Mahindra Global Chess League 2026 | Preliminary Stage — games with clocks and evaluations — Lichess
• Global Chess League 2026 Day 5: Sindarov Beats Firouzja To Give American Gambits Lead — Chess.com
• GCL 2026, Day 5: Gambits beat Kings, Pipers defeat Mumba Masters as race to the final heats up — FIDE

Related guides

• How to build an opening repertoire: 1.c4 as White and the Sicilian as Black
• An open games repertoire: meeting 1.e4 with 1...e5 without memorising Ruy Lopez theory
• How to avoid time trouble in chess: five rules for the last three minutes