Week 75: how to get better at the middlegame in chess, from five I threw away
Week 75 started with four losses and never really recovered. From June 30 to July 6, 2025, I played 10 rapid games, won 3, lost 7, and dropped 1280 → 1248 (−32). The losses had a common thread that matters for anyone asking how to get better at the middlegame in chess: in five of the seven I'd had a winning position first, and 22 of my 25 blunders came in the middlegame. My openings were fine. What happened after move 15 was not.
The week in numbers
| Games | 10 (4 White, 6 Black) |
| W / L / D | 3 / 7 / 0 |
| Rating | 1280 → 1248 (−32); low 1240 |
| Average accuracy | 72.3% |
| Blunders | 25 (my opponents: 24), 22 of them in the middlegame |
| Most played openings | Modern Defense with 1.e4 (3), Horwitz Defense (2), Indian Game (1), Pirc Main Line (1), Modern Defense with 1.d4 (1) |
Black was the problem colour: 1 win in 6. With White I split 2–2.
The five winning positions
Here are the best winning chances the engine gave me in the five lost games: 95.8% against Jokolono (1300), 86.9% against straker159 (1263), 84.6% against isakisob (1288), 84.4% against lubanzadio (1268) and 93.5% against TahsinKaanAl (1252). Three of those ended in checkmate against me, two I abandoned. Each of them was a middlegame where I was better and the game slipped away one move at a time.
Moment 1: a trade instead of a check
Game 1598 against Jokolono (1300), Black (game). White had castled queenside, my pieces were pointed at his king, and the engine had me at −3.35, winning.
Qa3+ is a check that also attacks two pawns in front of White's king: the queen slides from e7 to a3 along an empty diagonal. I chose to take a knight instead, hitting White's queen. It looked forceful, but White just recaptured with the queen and the evaluation dropped from −3.35 to +0.01. A trade gave White time; the check would have kept it from him. I lost this one 41 moves later by checkmate, with 28 seconds on my clock.
Moment 2: the knight that let go of the rook
Game 1599 against straker159 (1263), Black, a Pirc main line where the queens came off on moves 5 and 6 (game). My knight had been harassing White's rook: Nxc2, then Ra2, Nb4, Ra3. The engine had me at −5.09.
Nc2 keeps attacking the rook. Nd3 switched targets to the bishop on c1, and White answered 17.Nb5, which uncovered the rook's line along the third rank onto my knight. The evaluation went from −5.09 to −1.86. Two moves later I blundered again with 18...Nc5, and the game turned against me for good.
The game I lost in seven moves
The shortest loss of the week wasn't a middlegame at all. Game 1605 against jyoung1300 (1246) went 1.e4 g6 2.Nf3 d6 3.Bc4 Nf6 4.Ng5 e6 5.Qf3 d5 6.e5 Nfd7 7.Qxf7#. White's knight on g5 and queen on f3 were both aiming at f7, the weak square next to my king. My knight on f6 was the only piece blocking the queen, and when White's pawn attacked it, I moved it to d7. The engine marks 6...Nfd7 as the blunder: mate in one.
The lesson of the week
When you're winning a middlegame, the job is to keep the pressure, not to cash in early. In game 1598 a check against the castled king was stronger than a trade. In game 1599 renewing an attack on the same rook was stronger than switching to a new target. Both times the move I chose let my opponent breathe. A simple question before each move in a winning position: does this move force my opponent to respond, or does it give him a free move?
And from game 1605, a smaller one: before moving a piece that's blocking an attack, check what it was blocking.
All games this week
| Game # | Date | Colour | Opponent (rating) | Result | Opening | Accuracy | Link |
|---|---|---|---|---|---|---|---|
| 1597 | Jun 30 | White | falabutya (1302) | Loss | Indian Game | 67.8% | game |
| 1598 | Jun 30 | Black | Jokolono (1300) | Loss | Modern Defense with 1 e4 | 62.4% | game |
| 1599 | Jun 30 | Black | straker159 (1263) | Loss | Pirc Defense Main Line | 75.5% | game |
| 1600 | Jul 1 | Black | isakisob (1288) | Loss | Modern Defense with 1 e4 | 60.4% | game |
| 1601 | Jul 1 | White | boiledpotato4 (1244) | Win | Modern Defense with 1 d4 | 79.1% | game |
| 1602 | Jul 2 | White | lubanzadio (1268) | Loss | Queen's Pawn Opening Horwitz Defense | 77.4% | game |
| 1603 | Jul 2 | Black | TahsinKaanAl (1252) | Loss | Reti Opening Nimzo Larsen Variation | 65.0% | game |
| 1604 | Jul 5 | Black | Cmsin (1211) | Win | Modern Defense with 1 e4 | 92.6% | game |
| 1605 | Jul 5 | Black | jyoung1300 (1246) | Loss | Pirc Defense Modern Defense Geller System | 51.7% | game |
| 1606 | Jul 6 | White | ezzianebrahim (1246) | Win | Queen's Pawn Opening Horwitz Defense | 90.9% | game |
Previous: week 74 · Next: week 76 · Hub: the journey from 631 to 1385 · Find the moment your own winning middlegames slipped with AI chess analysis.
Frequently asked questions
How do you get better at the middlegame in chess?
From my week 75: when you're winning, look for checks and attacks that keep the pressure on before you trade or regroup. In two winning middlegames I traded or retreated when a check or a renewed attack kept the advantage.
Why do I keep losing winning middlegames?
In my case, by releasing the pressure. Five of my seven losses that week came after I'd had a winning chance of 84% or more, and 22 of my 25 blunders were in the middlegame.
Can you get checkmated in 7 moves?
Yes. In game 1605 it happened to me: 1.e4 g6 2.Nf3 d6 3.Bc4 Nf6 4.Ng5 e6 5.Qf3 d5 6.e5 Nfd7 7.Qxf7#. My knight left f6, and White's queen and knight both reached f7.