USA's round 4 opponents at the Chess Olympiad: how Ukraine and Spain play

Short answer

In round 4 of the 46th Chess Olympiad, on 19 September 2026, the United States beat Ukraine 3.5-0.5 in the open section and Spain 3.5-0.5 in the women's section. Using the classical games broadcast on Lichess over the last three years, this report shows each opponent's repertoire, the lines where each scores below their own average and where the engine finds the most mistakes, always with the sample size alongside.

Event: 46th FIDE Chess Olympiad

How this was compiled: Checked on 19 September 2026, with round 4 already finished: it started at 10:15 UTC. Collection of the classical games ended at 14:37 UTC and the error profiles were measured from 22:09 UTC. Pairings checked on Chess-Results; line-ups, ratings and titles from the official PGN broadcast on Lichess. The repertoire uses over-the-board classical games from the last three years broadcast on Lichess, matched to each player by FIDE ID and excluding round 4 itself: 102, 158, 101 and 75 games for the four Ukrainians in the line-up and 110, 128, 100 and 113 for the four Spaniards. Error profiles measured with Stockfish 19 at depth 16 on each player's 12 most recent classical games. Online accounts are not included in this version. Final round 4 scores checked against the official PGN at 22:35 UTC.

The essentials in a few lines

• In round 4 of the Samarkand Olympiad, on 19 September 2026, the United States beat Ukraine 3.5-0.5 on match 44 of the open section and Spain 3.5-0.5 on match 12 of the women's section.
• Ukraine fielded four players with samples of 75 to 158 broadcast classical games — the deepest opponent dossier of the US campaign so far.
• Anton Korobov scores 70% with White in 10 games of the Italian with 3.Bc4, but 50% in the 4 games that continue 3...Bc5. Wesley So played 3...Bc5 and won.
• Ihor Samunenkov's weakest line as Black is his own main defence: 46% in 13 games of the Petrov, against 59% for his own Black average.
• Spain's Marta Garcia Martin has scored 33% in 3 games against the King's Indian with White; Carissa Yip played the King's Indian and won.
• All figures are from the opponent's point of view and show tendencies: samples range from 75 to 158 games in the open section and 86 to 128 in the women's section.

Round 4 at a glance

The United States were higher rated on all eight boards, but by the narrowest margins of their campaign so far: 117 to 217 points in the open section, 90 to 166 in the women's. Ukraine rested Igor Kovalenko, at 2642 the highest-rated player of their squad, and still fielded four players above 2500.

The sample column is the deepest the series has seen. Anton Korobov has 158 broadcast classical games in the last three years, Vasyl Ivanchuk 102, Ihor Samunenkov 101 and Roman Dehtiarov 75. On the Spanish side the four players in the line-up have between 100 and 128. Nothing here rests on three or four games, except where the text says so.

Depth of sample is not the same as depth of information. Ivanchuk drew 32 of his 52 games with Black in the sample — a 62% score built almost entirely out of draws, which tells you how the games end but very little about where they go wrong.

Ukraine: where the big samples point

Anton Korobov, board 2, has the largest sample of the series so far, and it is detailed enough to separate the trunk from the branch. With White he opens 1.e4 in 46 of 80 games and scores 71%; against 1...e5 he reaches the Italian with 3.Bc4 in 10 games and scores 70%. But the four games that continue 3...Bc5 are a different story: 50%, one win, two draws and one loss. With Black the clearest hole is the French — 29% in 7 games, against his own 62% average — and the Sicilian with 2...e6 3.d4 at 50% in 7.

Ihor Samunenkov, board 3, has the one weak line in this round that rests on a sample worth the name. With Black he scores 59% overall, but 46% in 13 games of the Petrov, 1.e4 e5 2.Nf3 Nf6, which is his own most played defence. The King's Knight Opening with 2...Nc6 is 42% in 6. What he does not see often is 1.c4: three games in the sample, and he scored 83% in them.

Roman Dehtiarov, board 4 and the lowest rated of the four, is the most lopsided: 76% with White in 35 games against 56% with Black in 40. The Black hole is precise — 46% in 13 games after 1.e4 e5 2.Nf3 Nc6. As White his Sicilian sample is tiny: five games, scoring 90%, and none of them reached the Rossolimo with 2...Nc6.

Vasyl Ivanchuk, board 1, is the hardest of the four to scout, and the reason is in the results column. With Black he scored 62% in 52 games with 16 wins, 32 draws and 4 losses. The lines where he drops below that average all rest on three or four games: 33% in 3 games after 1.e4 e5 2.Nf3 Nc6 3.Bc4, 33% in 3 in the French, 50% in 4 in the King's Indian. The dossier marks all of them low confidence, and so does this text.

Spain: 1.e4 without exception, and a Rossolimo problem

Sabrina Vega Gutierrez, board 2, is the most predictable opponent of the round and the best documented: 1.e4 in all 64 games with White, 128 games in the sample. After 1.e4 c5 2.Nf3 Nc6 she scored 37% in 15 games, against 62% for her own White average — the only weak line of this round that the dossier marks as high confidence. Against that 1.e4, opponents answered 1...c5 in 36 of the 64 games, and there she scored 54%, eight points below her own White average.

Marta Garcia Martin, board 1, spreads her first move: 1.e4 in 28 games, 1.d4 in 17, 1.c4 in 10, 1.Nf3 in 5. The line where she drops is the King's Indian — 33% in 3 games after 1.d4 Nf6 2.c4 g6, against 61% overall. Three games, and the dossier says low confidence. With Black the split is sharper and better documented: 78% in 23 games against 1.e4, 47% in 16 against 1.d4.

Monica Calzetta Ruiz, board 4, is the one whose numbers depend on the move order. After 1.e4 c5 2.Nf3 Nc6 she scored 69% in 8 games. But when the fianchetto comes early — 1.e4 c5 2.Nf3 g6 — she scored 22% in 9, with one win, two draws and six losses. Same pawn structure, opposite results, and the difference is which move order gets there.

Ann Matnadze, board 3, is the hardest to pin down: 69% with White in 51 games, 63% with Black in 49, and no weak line above four games in either colour. She is a 1.d4 player in 32 of 51 games, and in the 10 that continue 1...d5 she played 2.c4 four times, 2.Nf3 four times and 2.g3 twice. Ines Prado Acebo, who rested, is the Spaniard the dossier describes best: 31% in 16 games as White after 1.d4 Nf6 2.c4 e6.

Where the engine finds the mistakes

The table shows the average loss per move, the ACPL, in each phase. Ukraine's figures are the lowest the series has measured: Vasyl Ivanchuk sits at 13 centipawns overall, with 4 in the opening, and the other three are between 17 and 19. That is what four players above 2500 should look like, and it is also why an opening dossier says more about them than an error profile does.

With 12 games per player, ACPL is good for comparing the phases of one player, not for ranking players against each other. Three centipawns between Samunenkov and Korobov mean nothing. The contrast between phases within one name does: Anton Korobov loses 26 centipawns per move in the middlegame and 11 in the endgame; Monica Calzetta Ruiz draws the same shape more sharply, with 46 in the middlegame and 13 in the endgame.

The highest figure in the table is Sabrina Vega Gutierrez's endgame: 61 centipawns per move, over 202 measured endgame moves, against 34 in the middlegame. The move count there is large enough to describe a tendency, which is not the case for Ihor Samunenkov's 40 in the same column: his 12 games produced only 37 endgame moves, and nobody is described by that.

The blunder count is left out. When a player who is already winning picks a slower mate, the blunder rule records a huge loss for a move that changes nothing. ACPL caps each evaluation at a thousand centipawns and does not suffer from that.

What round 4 showed

Both matches finished 3.5-0.5, and the openings were not an accident. Wesley So answered Korobov's Italian with 3...Bc5 and won on board 2: in the sample, Korobov scored 70% in the 10 games that reached 3.Bc4 and 50% in the 4 that continued 3...Bc5. Carissa Yip met 1.Nf3 with a King's Indian set-up and won on board 1, the structure in which Marta Garcia Martin scored 33% — in three games, which is the honest way to write it.

Tatev Abrahamyan's win is the most interesting of the four, because it did not follow the line but the structure. Calzetta answered 1.e4 with 1...c5 2.Nf3 Nc6, where the dossier gives her 69% in 8 games. Abrahamyan played 3.Bb5 and then let her fianchetto — reaching the position type in which Calzetta scored 22% in 9 games, by a move order the dossier files elsewhere. A repertoire tree is organised by moves; the trouble a player has is organised by positions, and the two do not always line up.

Caruana's 3.Bc4 against Ivanchuk is the one to be careful with. Ivanchuk scored 33% in that line — three games, one loss and two draws, a sample that would not survive one more result in either direction. Caruana won, and that is one data point, not a confirmation. The same caution applies in reverse to Hans Niemann's 1.c4 against Samunenkov: the dossier said Samunenkov had scored 83% in his three games against 1.c4 and 46% in thirteen games of the Petrov, and the game was drawn in 21 moves.

The one board where the dossier had little to offer was Levon Aronian's. Roman Dehtiarov opened 1.e4 and met 1...c5 with the Rossolimo — five Sicilians in his White sample, all scoring 90%, and none of them with 2...Nc6. Aronian won anyway. A dossier that does not cover a position is not a dossier that says the position is bad; it says nobody has tested it, which is a different sentence and a much shorter one.

Frequently asked questions

Update log

• 2026-09-19 — First published after round 4 finished: repertoire from the classical games broadcast on Lichess, error profiles with Stockfish 19 at depth 16, and final scores checked against the official PGN.
• 2026-09-29 — The Olympiad ended on 27 September: Uzbekistan won the open and China the women's event. This coverage is no longer updated; the final result is in the standings story.

Sources

• 46th FIDE Chess Olympiad Samarkand 2026 | Open — round 4, matches 1-12 (PGN with line-ups) — Lichess
• 46th FIDE Chess Olympiad Samarkand 2026 | Women — round 4, matches 1-25 (PGN with line-ups) — Lichess
• Anton Korobov — broadcasts under FIDE ID 14105730 — Lichess
• 46th Chess Olympiad Samarkand 2026, Women — round 4 pairings — Chess-Results

Related guides

• How to build an opening repertoire: 1.c4 as White and the Sicilian as Black
• Corporate chess leagues: what changes when the point belongs to the team
• A 7-day chess tournament preparation plan: from repertoire to energy management