Back rank mate in a winning position: one rook move in game 1768
Game 1768 ended with a back rank mate in a winning position. Before my 25th move I was +7.01, a 93.0% winning chance, the highest of the game. I moved my rook from e5 to h5. Black's queen checked on c1, my only legal reply was to block with my queen on d1, and the queen capture on d1 would have been mate. The record says the game was abandoned there, a loss.
The recorded game
I played this ten-minute rapid game (no increment) on October 31, 2025 as White against Olsmokey1 (1254). My recorded rating for it was 1253. The opening went badly, with my bishop lost for a pawn by move 6; that part, and the Horwitz Defense with an early g5 that caused it, is on the opening page for this game. This page starts at move 13, when I was winning.
| Detail | Saved record |
|---|---|
| Game | 1768 of the From 631 to 1385 journey |
| Result | Loss (0-1), "Olsmokey1 won - game abandoned" |
| My accuracy | 44.1% (Black 56.7%) |
| My blunders / mistakes / inaccuracies | 8 / 2 / 1 |
| Black's blunders / mistakes / inaccuracies | 7 / 0 / 3 |
| Middlegame accuracy | 42.4% over 13 moves |
| My winning chance before moves 13 / 20 / 25 | 90.2% / 86.8% / 93.0% |
| My clock after move 24 / move 25 | 384 s / 335 s |
| Original score | Chess.com game |
I came into this game on three wins in a row (1765 to 1767). It was a strong week: the week 92 diary shows 1202 to 1301 over 43 games. This loss was one of the 15.
Position 1: move 13, the queen chase instead of the knight
After a wild opening, Black's king had walked to f8 (12...Kf8, a blunder) and the evaluation was +6.03. My queen was on h5, Black's queen on a5.
I gave a queen check on f3 that also attacked the knight on d5. Today's check rates it "clearly better for white" (+2.57), but the saved evaluation after it was +0.74. The engine's first choice was the knight to c4, attacking Black's undefended queen, rated "white is completely winning" (+7.26). The line continues with the knight jumping to d6, where Black gives the queen for it.
It was the third of four moves in a row with the same engine suggestion. On moves 11, 12, 13 and 14 Stockfish wanted Nc4, and each time I moved my queen or my bishop instead. The moves themselves show the habit: I kept choosing checks and captures over a knight move.
Position 2: move 20, the pawn that chased the queen
Black gave me the advantage back with 17...b5. My bishop took a knight on d5 (the review marks 18.Bxd5 as a good moment), my queen took the other knight on d4, and by move 20 I was +5.12.
I attacked Black's queen with a pawn. The engine's choice was the quiet knight development 20.Nc3, "white is winning" in today's check. My pawn move scored +1.57 in the saved analysis. Look at that knight on b1: it had not moved in the whole game. Stockfish wanted it out on move 20 and again on move 21, when I played the other knight to f3 instead. That undeveloped knight matters in the last position.
Position 3: move 25, the rook leaves the e-file
Black blundered again (20...Qc7), we traded on e5, and after 24.Rxe5+ Kd6 my rook stood on e5, next to Black's king on d6. The evaluation was +7.01. I thought for about 48 seconds, the longest think of my game.
My rook went to h5. The checker describes it only as a rook move; the record doesn't say what I was planning, and I won't guess. The saved evaluation after it is mate in two for Black. Stockfish's saved choice was 25.Na3. Today's check ranks a queen move to f4 first, then Na3 and Nc3, all "white is completely winning".
What those three moves share became clear when I ran one more check. I gave the same position to the checker with Black to move, as if I had passed. Its first move is the queen check on c1, mate in three: 25...Qc1+ 26.Qd1 Qxd1+ 27.Re1 Qxe1, checkmate. So Black already had a threat against my first rank, and I needed a move that dealt with it. The queen on f4 covers c1 along the diagonal. The knight to a3 clears b1, so my rook on a1 guards the rank. The knight to c3 blocks the c-file. My rook move did none of that.
Position 4: the mate
The c-file was open from c7 to c1. Black's queen went all the way down with check. My king on g1 had pawns on f2, g2 and h2 in front of it, so it had no square to step to. The checker's line is 26.Qd1 Qxd1, checkmate, and the position after the check has exactly one legal move for me: the queen to d1. My knight on b1 couldn't help, and my rook on a1 was stuck behind it.
A back rank mate happens when the king is shut in by its own pawns and no piece covers the first rank. Here the cover was missing before my move, because the knight on b1 blocked my rook on a1. With the rook still on e5 it could at least drop back to e1 and delay the mate by one move; on h5 it couldn't even do that.
The lesson of game 1768
Three positions, one pattern. On moves 13, 20 and 25 I was clearly winning, and each time I chose the active move: a queen check, a pawn attack on the queen, a rook swing. The engine wanted the slow moves: a knight to c4, the knight from b1 out to c3, the knight to a3. The knight on b1 never moved in 25 moves, and in the final position it was the piece that kept my rook on a1 from guarding the first rank.
The question I skipped on move 25 was the simplest one: what does my opponent threaten? Black's queen on c7 looked down an open c-file at c1, a square on my first rank that nothing of mine guarded. One check there was enough. When my king sits behind three unmoved pawns, I now look at every open file and diagonal that reaches my first rank before I play an attacking move. For more on this pattern, I have an answer on why winning positions get lost, and an older back-rank story from game 412.
What changed afterwards
The record shows I won the next game, 1769, with the same opening, then lost 1770 and won three in a row after that (1771 to 1773). It doesn't show what I thought of this loss, and I won't invent it. If you want to see your own key moments this way, with arrows for the move played and the better one, Chess Coach works in Claude and ChatGPT. The guide to reading Stockfish analysis explains what "mate in two" means on the evaluation bar, and the journey hub has the rest of the series.
Frequently asked questions
What is a back rank mate?
A checkmate on the first rank (the eighth for Black), when the king is shut in by its own pawns and no piece can block or capture the checking piece. In game 1768 it came with a queen check on c1 and a queen capture on d1.
How winning was I before the mistake?
The saved evaluation before my 25th move was +7.01, and my winning chance 93.0%, the highest of the game. After 25.Rh5 the engine shows mate in two for Black.
Which move would have kept the win?
The saved analysis gives 25.Na3, a knight move that frees my rook on a1 to guard the first rank. Today's check also rates 25.Qf4, covering c1, and 25.Nc3, blocking the c-file, as completely winning for White. All three deal with Black's threat of a queen check on c1.
How did the game end?
After 25...Qc1+ I had one legal move, the queen to d1, and then Qxd1 was mate. The record says Olsmokey1 won and that the game was abandoned at that point.