Pawn storms with opposite castling · For club players · 18 min read
Opposite-Side Castling: Giri's Pawn Storm Race, Checked
Giri beat Maurizzi in 27 moves at the 2025 Grand Swiss. Stockfish says White was winning at move 14, and shows how a pawn-storm race is really lost: one tempo at a time.
- plan
- target
- White's storm
- Black's storm
- White's king
In short
- Opposite-side castling turns a game into a race: each side pushes pawns at the other king, and the first storm to open a file usually wins.
- In Maurizzi v Giri (Grand Swiss 2025) White stormed on the kingside while Black's king was still on e8, and Black stormed on the queenside against a king on c1.
- Giri won in 27 moves, but Stockfish says White was clearly better after 13...Na7 and gave that away with one slow move (14.Bg2), then lost the race with 19.f5 and 22.Bh3.
- The lessons are practical: count tempi before you push, trade the pieces that guard the enemy king, and do not rush your own king to a side until the storm has shown itself.
How we know what we say
- Verified From the record: Stockfish's numbers, the players' own notes or interviews, or a cited source.
- Inferred Read from the moves and the position: a reasoned reconstruction, not the player's words.
- Speculative A guess about mood, intent or practical motives: plausible, not proven.
Part 1Samarkand 2025: a Najdorf and a race
The FIDE Grand Swiss is an 11-round open tournament whose winner and runner-up earn places in the Candidates tournament. The 2025 edition was played in Samarkand from 3 to 15 September with 116 players, and Anish Giri won it with 8 points and qualified for the 2026 Candidates. [5] Verified
In round 6 Giri, rated 2746, had Black against the French grandmaster Marc'Andria Maurizzi (2610), who had beaten three strong players in the three rounds before and arrived with momentum. The win put Giri among the leaders, half a point behind Parham Maghsoodloo. [3][6][7] Verified
| White | Black | |
|---|---|---|
| Player | Marc'Andria Maurizzi (France) | Anish Giri (Netherlands) |
| Rating, September 2025 | 2610 | 2746 |
| Result | Resigned after 27 moves | Won, then won the tournament |
| Opening | Najdorf Sicilian, 6.Nb3 and the g4 storm | e6 and ...b5-b4, king left in the centre |
The game is short and it looks like a clean win for the faster attacker. It is a better lesson than that, because a race is not decided by who pushes first but by who loses a tempo last.
Part 2What is opposite-side castling, and who wins the race?
Opposite-side castling means the two kings live on different wings, so each side can throw pawns forward at the enemy king without weakening its own. The pawns act as battering rams: they are there to open a file or a diagonal for the rooks and queen, not to win material. Inferred
The race has simple arithmetic. Count how many pawn moves each side needs to open a line against the king, add the moves needed to bring the pieces up behind them, and compare. The side with fewer moves left usually wins, which is why a single wasted move, a quiet developing move at the wrong time, can swing the whole game.
- plan
- target
- The storm
- plan
- target
- The storm
The same plan with a different number of moves left. The side with fewer moves to open a file is winning the race.
A race is not only pawns. Pieces that already stand on the right squares save moves, and exchanges that remove the enemy king's defenders (a knight on the king's wing, a bishop on a key diagonal) shorten the count without a single pawn push. Watch for both in the game.
Part 3How the opening created the race
Maurizzi chose an unusual path in the Najdorf Sicilian: 6.Nb3, then Be3 and the g-pawn rush g4-g5. Giri answered with ...e6, ...Nc6 and ...b5-b4, which attacks the knight on c3 and starts a queenside storm of Black's own. Neither king had castled, so each side was building a storm against a king that had not yet chosen its home.
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Nb3 e6 7.Be3 Nc6 8.g4 b5 9.g5 b4 10.Na4 Nd7 11.Qe2 Be7 12.O-O-O Bb7 +0.75White slightly better
White castles long on move 12. The pawn on g5 has driven the knight from f6, and Black's pawn on b4 has driven the knight to a4. Verified
- plan
- key square
- White's g-pawn
- Black's b-pawn
The kings tell the story. White castled queenside on move 12, so the storm that Giri had started with ...b5-b4 now had a target. Black did not castle until move 25, long after White's kingside pawns had run out of steam: a rare case where the defender's king waited in the centre and was never really in danger. Inferred
Part 4Two plans from the same position
After 12...Bb7 the position is a symmetrical race. White has the g-pawn on g5 and wants f4-f5 to open lines towards a king that may still go either way. Black has b4 and a6 and wants ...a5-a4, with the bishop on b7 and queen ready to join, against a king on c1 whose shelter has already been loosened by the knight on a4.
- plan
- target
- White's grip
- plan
- target
- Black's storm
From the same position: White wants a file open on the kingside, Black wants one open on the queenside. The next few moves decide whose is shorter.
In this game Black's defence and attack run on the same pawns: the b4 and a-pawn restrict White's king and are also the lever. White's pieces have more to do (queen, rooks, bishops), because the king on c1 is where Black's attack lands.
Part 5The decisions that mattered
Three decisions turned the race, and each is a tempo question. Try the problems before the answers; the number chips come from Stockfish 19.
Critical moment 1 of 4 · Move 13, Black to play
A knight reroute that leaves a pawn loose
- threat
- key square
- target
- Black's pawns
White has just played f4. Black's queenside storm needs b4 to stay protected, and the black knight on c6 is guarding it. Which move continues the plan without leaving anything loose?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 13...Na5 14.Bd2 Bc6 15.Bxb4 Bb5 | +0.06about equal | Best: close to level, so Black keeps the queenside storm going with the knight still guarding b4. Verified |
| 13...O-O 14.Qd2 a5 15.Rg1 Qc7 | +0.87White slightly better | Playable: White is better by about four fifths of a pawn. Verified |
| 13...Qc7 14.Qd2 Rb8 15.Kb1 Ba8 | +0.78White slightly better | Playable: White is better by about four fifths of a pawn, the same as castling. Verified |
What Giri played
13...Na7+1.21White clearly better
Giri plays 13...Na7, moving the knight away from c6 to prepare ...Bxa4 and the exchange of White's knights.
Stockfish preferred 13...Na5 (+0.05); the game move gave up about 1.2 pawns. Depth 25.
What it revealed
The knight on c6 was the only guard of b4 and a blocker in front of the bishop on b7. Moving it left b4 loose and the e4 pawn attackable, and only one reply, 14.Qc4, handles both.
What Stockfish adds Stockfish prefers 13...Na5, which keeps a knight guarding b4 and is level. 13...Na7 is playable only because 14.Qc4 is a hard move to find.
- Verified Stockfish rates 13...Na7 as an inaccuracy of about a pawn and a half: the position was close to level with 13...Na5, and after the knight move White is better by about a pawn and a half, but only 14.Qc4 keeps it.
- Inferred The idea of ...Na7 is to trade the knight on a4 and so rip open the white king's cover with ...Bxa4. It works in the game, but only because White does not find the one good reply.
- Speculative Giri may have judged that the queenside exchanges were worth a little risk. The game gives no sign of the real reasoning.
Critical moment 2 of 4 · Move 14, White to play
Find the one move that does two jobs
- threat
- target
- White's king side
Black's last move left the b4 pawn without a guard and lets the bishop on b7 attack the e4 pawn, which the queen on e2 cannot guard because the bishop on e3 stands in the way. What does White play?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 14.Qc4 14...O-O 15.Qxb4 Qc7 16.h4 | +1.51White clearly better | Best: keeps about a pawn and a third for White, because the queen guards e4 and attacks b4 at once. Verified |
| 14.Qd2 14...Bxe4 15.Bg2 Bxg2 16.Qxg2 | −0.76Black slightly better | A mistake: 14...Bxe4 wins a pawn and Black is better by about four fifths of a pawn. Verified |
What Maurizzi played
14.Bg2−0.19about equal
Maurizzi plays 14.Bg2, adding a second guard to e4.
Stockfish preferred 14.Qc4 (+1.21); the game move gave up about 1.4 pawns. Depth 25.
What it revealed
The bishop move defends e4 and develops a piece, which is natural in a race. It does nothing about b4 and leaves Black time for ...Bc6, hitting the knight on a4. The race tilts to Black.
What Stockfish adds The usual verdict on this game blames the later pawn push 19.f5. Stockfish finds the decisive tempo five moves earlier: after 13...Na7 White was a pawn and a half better, and 14.Bg2 is where the race changed hands.
- Verified Stockfish: 14.Qc4 keeps about a pawn and a third for White; 14.Bg2 gives that back and Black is better by about a third of a pawn after 14...Bc6.
- Inferred Qc4 does two jobs: it guards e4 along the fourth rank and attacks b4, so it solves both of Black's last move's problems at once. A developing move can only solve one.
- weakness
- target
- White's king
- pawn chain or line of force
The exchanges in the middle are the quiet heart of the plan: Giri trades the bishop on b7 for two knights (the one on a4 and the one on b3), and White retakes with the a-pawn. The two knights were the pieces that guarded the white king's flank, and the pawn structure that remains, b2, b3 and c2, is brittle. The open a-file points straight at the king. Inferred
Critical moment 3 of 4 · Move 19, White to play
Push on, or run?
- threat
- target
- White's king
Black has just played ...a5 and threatens ...a4, with the queen already aimed at c2. White's pawns on f4 and g5 are ready for a break in front of the black king, still on e8. Is it time to push, or time to do something about the king on c1?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 19.Rd4 19...O-O 20.Rc4 Qb7 21.Qd2 | −0.43Black slightly better | Best: the rook hits b4, the pawn that leads Black's storm, and White stays within about two fifths of a pawn of equality. Verified |
| 19.Kd2 19...O-O 20.Ra1 Nc5 21.Rhd1 | −0.81Black slightly better | Good: the king leaves the a-file and White stays within about four fifths of a pawn of equality. Verified |
| 19.Qe2 19...O-O 20.Rd4 Nb6 21.Kd2 | −1.30Black clearly better | Holds: Black is better by about a pawn and a third, with the queen covering the c2 square. Verified |
| 19.Rd2 19...a4 20.Kd1 h6 21.gxh6 | −1.63Black clearly better | Worse: the rook leaves the fourth rank and Black is better by about one and two thirds pawns. Verified |
What Maurizzi played
19.f5−2.53Black winning
Maurizzi plays 19.f5, pushing the storm forward.
Stockfish preferred 19.Rd4 (−0.42); the game move gave up about 2.1 pawns. Depth 24.
What it revealed
The push is a pawn move that does not shorten White's count: Black answers 19...a4 at once, and White's break on e6 opens the f-file for Black's rook as much as for White's pieces.
What Stockfish adds The best defence is a counter-strike at the spearhead: the rook hits b4, the pawn that leads Black's storm. It is the kind of move a storming side finds hard to make, because it spends a tempo on defence when the attack still feels live.
- Verified ChessBase India marks 19.f5 with a question mark and says it was not a good idea because Black immediately launched an attack of its own.[4]
- Verified Stockfish agrees: 19.Rd4, 19.Kd2 and 19.Qe2 keep Black's edge to about a pawn and a third or less, while 19.f5 leaves Black better by about two and a half pawns.
Part 6The turning point: 22.Bh3 and the b3 break
After 19.f5 a4 20.fxe6 fxe6 21.bxa4 Rxa4 White is already close to three pawns worse by Stockfish's count. The rook on a4 is a dominant piece and the f-file has opened for Black. One more tempo error ends the game.
Critical moment 4 of 4 · Move 22, White to play
The last chance to slow the storm
- threat
- target
- key square
- White's king
Black's rook is on a4, a pawn stands on b4 and the queen eyes c2. White's king has only the pawns on b2 and c2. White has a free move and a loose pawn to hit on e6. What is the most useful move?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 22.Qe2 22...b3 23.c3 O-O 24.Kd2 | −3.05Black winning | The most resistant, though Black is still better by about three pawns. Verified |
| 22.Rd2 22...Ra1 | −#1Black mates by force | A blunder: 22...Ra1 is checkmate. Verified |
What Maurizzi played
22.Bh3−6.02Black completely winning
Maurizzi plays 22.Bh3, attacking the loose pawn on e6.
Stockfish preferred 22.Qe2 (−3.30); the game move gave up about 2.7 pawns. Depth 20.
What it revealed
Taking a pawn on e6 is a move that does nothing about Black's attack. Black replies 22...b3 and the pawn hits c2, with the rook on the open a-file and the queen behind.
What Stockfish adds Even the best defence, 22.Qe2, leaves White about three pawns worse. 22.Bh3 turns a bad game into a lost one, with Black more than six pawns ahead. Black's reply 22...b3 is Stockfish's own choice.
- Verified Chessdom criticises 22.Bh3 as ignoring Black's main threat and describes 22...b3 as leaving White's king under fire.[2]
- Verified After 22.Bh3 Stockfish puts Black ahead by more than six pawns, against about three after the best defence 22.Qe2.
22...b3 23.c3 Nc5 24.Qe2 Qc6 25.Kd2 O-O 26.Rhf1 Rxf1 27.Qxf1 Qxe4 −6.65Black completely winning
The game ends: Black's knight arrives on c5, the king castles at last on move 25 and 27...Qxe4 wins a pawn with the white king exposed. White resigned. Verified
- plan
- target
- outpost
- White's king
Part 7What White could have done: step away from the storm
The game was lost at move 19, and it was still playable. After 18...a5 Black is only slightly better, and the white king on c1 is the target of the whole Black army. Stockfish's answer is to strike at the spearhead of the storm: 19.Rd4 attacks the b4 pawn, and if Black defends, White has gained a tempo to bring the king to safety.
- idea not played
- target
- key square
- White's king
19.Rd4 O-O 20.Rc4 Qb7 21.Qd2 Rfb8 22.Kd1 −0.54Black slightly better
Stockfish's line: the rook goes to d4 and c4, the queen to d2 and the king to d1, and White stays within a pawn of equality. Verified
Part 8A fresh angle: the race was lost at move 14, not 19
| Move | Who | Before | After |
|---|---|---|---|
| 13.f4 | Maurizzi | about three quarters of a pawn for White (13.Qd2) | close to level |
| 13...Na7 | Giri | close to level (13...Na5) | about a pawn and a half for White |
| 14.Bg2 | Maurizzi | about a pawn and a third for White (14.Qc4) | about a third of a pawn for Black |
| 19.f5 | Maurizzi | about two fifths of a pawn for Black (19.Rd4) | about two and a half pawns for Black |
| 22.Bh3 | Maurizzi | about three pawns for Black (22.Qe2) | more than six pawns for Black |
Reports describe the same game differently: the ChessBase and Chess.com round reports speak of a direct kingside attack, while Chessdom and the board agree on the queenside. The pawns that moved were on the a- and b-files and the king that was hit was on c1. This is a small reminder that a game summary is not the game. [2][6][7] Verified
The second surprise is the king that did not castle. Black's king sat on e8 for 24 moves, through two enemy pawn pushes on its own wing, and was never in real danger because White's kingside pawns had nothing behind them. Waiting to castle until the attacker commits is a practical weapon in the Sicilian. Inferred
Part 9How to win a pawn-storm race in your own games
The recipe
- Count the race firstBefore you push a pawn, count the moves each side needs to open a file against the king. Push only if your count is shorter or equal and you are to move.
- Trade the defendersGiri swapped a bishop for two knights, and the exchanges ripped the cover from the white king. Look for trades that remove the pieces guarding the enemy king, not just the ones that win material.
- Keep your own king flexibleBlack's king stayed on e8 until move 25. Do not castle into a storm you can see coming; wait for the attacker to commit pawns that cannot go back.
- Make every move shorten your countA developing move that does not threaten anything, like 14.Bg2, can hand the tempo to the other side. In a race, choose the move that solves two problems at once.
- If you are slower, stop racingHit the pawn that leads the storm, step the king off the line, or change the structure. Do not add pawn pushes that cannot arrive first.
- Count tempi before pushing a pawn in front of your own king.
- Trade the pieces that stand beside the enemy king.
- Find the move that does two jobs; a single-purpose move loses a tempo.
- Open a file with a trade or a lever, not with a pawn that has nothing behind it.
- If you are behind in the race, hit the leading pawn or move the king; do not add pawn pushes.
Your turnPractise it
Set each position up, decide your plan before you look, then play it out against Stockfish to see whether it holds.
White to play, move 13. White has castled long, and Black has not castled. Which queen move keeps the most advantage?
Hint
Which queen move eyes the b4 pawn and stays useful to White's kingside pawns?
A good answer
13.Qd2. Stockfish puts White about three quarters of a pawn ahead. The game move 13.f4 lets Black equalise with 13...Na5.
White to play, move 14. Black has just moved the knight away from c6. Find the only move that keeps the advantage.
Hint
Two things are loose: a pawn on b4 and the e4 pawn. One queen move covers both.
A good answer
14.Qc4. It guards e4 along the fourth rank and attacks the unguarded b4 pawn. White keeps about a pawn and a third; 14.Qd2 loses a pawn to ...Bxe4.
White to play, move 19. Black threatens ...a4, and the black king is still on e8. Is the pawn push f5 the right idea?
Hint
Count the moves each side needs to open a file. Black is one move from opening the a-file. Is there a way to hit the spearhead?
A good answer
No. No. 19.Rd4 is Stockfish's choice: the rook hits b4 and White stays within about two fifths of a pawn of equality. 19.Kd2 is also fine, while 19.f5 gives Black about two and a half pawns after 19...a4.
White to play, move 22. White is already close to three pawns worse. Which move gives the best chance to survive: 22.Bh3, 22.Rd2 or 22.Qe2?
Hint
Black's rook on a4 and queen on c7 threaten the king. Check every move for a back-rank trick before you play it.
A good answer
22.Qe2 is the most resistant, though White is still about three pawns worse. 22.Rd2 loses to 22...Ra1 mate, and 22.Bh3 lets Black push 22...b3 with a winning attack.
Part 10What to take away
- A race is a tempo count. The faster storm wins; the slower one loses even if each push looks natural.
- One slow move is enough. 14.Bg2 gave back a pawn and a half and the game never recovered.
- Trade the defenders. Three exchanges tore the cover from the white king before a single pawn reached it.
- A king can wait. Black castled on move 25 and was never in danger because the attack against it stalled.
- When behind, stop racing. 19.Rd4, a strike at the leading pawn, kept the game alive.
Questions readers ask
What is opposite-side castling in chess?
Opposite-side castling means the two kings castle on different wings, one short and one long. Each side can then push pawns at the enemy king without weakening its own, which usually turns the game into a race to open files against the kings.
Who wins a pawn-storm race in chess?
The side that needs fewer moves to open a file against the enemy king usually wins. Count the pawn pushes each side still needs, add the moves to bring the pieces up behind them, and avoid any move that does not shorten your count, as Maurizzi v Giri (2025) shows.
When should I castle on the opposite side from my opponent?
Castle on the opposite side when you have a faster attack than your opponent or when your king's home is safer than theirs. In the game Giri's king waited on e8 until move 25, because castling early would only have given White's pawn storm a target.
How do I defend against a pawn storm?
Trade the attacker's most dangerous piece, keep pieces beside your king, and if you are slower in the race, strike at the pawn that leads the storm or step the king away from the line of fire. In the game 19.Rd4, a rook strike at the leading b4 pawn, was Stockfish's defence, while the push 19.f5 lost the race.
Which wing did Giri attack in Maurizzi v Giri 2025?
Giri attacked on the queenside, with the b- and a-pawns, against a white king on c1. Some reports call it a kingside attack, but the pawns that moved were on the a- and b-files and the pieces that landed were on the a-file and the c2 square.
Sources and method
The game score was checked against two independent sources. Every position, line and candidate in this article was replayed for legality, and every engine number was produced by Stockfish 19 WASM Multithreaded for this article (depth 19 to 40, deepest at the critical moments), not copied from elsewhere. What the players thought is reconstructed from the moves, their published notes and later analysis; it is labelled as verified, inferred or speculative, and nothing is presented as a quotation unless a source is cited for it.
- Maurizzi v Giri, FIDE Grand Swiss 2025 round 6: official game score from the Lichess broadcast
- Chessdom: Anish Giri leaps to shared second, eyeing Candidates 2026 spot (moves 19 to 27, commentary on 22.Bh3 and 22...b3)
- FIDE: Grand Swiss 2025 round 6 report (moves 19 to 27, result, standings)
- ChessBase India: Grand Swiss 2025 round 6 report (moves 19 to 27, 19.f5 marked as an error, ratings)
- Wikipedia: FIDE Grand Swiss Tournament 2025 (dates, venue, time control, qualifiers)
- ChessBase: Grand Swiss round 6 report, analysis by Andre Schulz
- Chess.com blog: Grand Swiss round 6 (Maurizzi's run of wins before the game)
Read next
Morphy's famous 1858 win against two amateurs, and what Stockfish 19 says about when the uncastled king was really lost.
Active King in the Endgame: Capablanca's King Walk, CheckedCapablanca gave up two pawns to march the king into Tartakower's camp in 1924. Stockfish 19 says the king walk was sound, but the game was decided two moves earlier.
Pawn Breaks in Chess: Fischer's e4 and f5 v Spassky, CheckedFischer's 20.e4 is the most praised move of Reykjavik 1972, but Stockfish 19 says it only asked a question. Learn when a pawn break wins and when it merely invites a mistake.
Turn reading into skill
Strategy sticks when you use it. Practise real positions with Stockfish's verdict at once, or play the plans out from any diagram above.