Week 116: why do I keep blundering pieces? Three captures that backfired
The week of April 27 to May 3, 2026 ended where it started: 8 rapid games, 1205 → 1206, four wins and four losses. It started badly. My first game was a loss that dropped me to 1197, one of only two games between April and July that left me below 1200. I climbed above 1210 three times and fell back each time. Looking at the engine's turning points, I can give a concrete answer to a question every improver asks, why do I keep blundering pieces: almost always, it happens right around a capture.
The week in numbers
| Games | 8 |
| Wins / losses / draws | 4 / 4 / 0 |
| Rating | 1205 → 1206 (+1), lowest point 1197 |
| Average accuracy | 78.1% |
| My blunders | 17 |
| With White / with Black | 2–3 / 2–1 |
| Most played | Queen's Pawn Opening, Zukertort Chigorin Variation (2); six other openings once |
Eight games and seven different openings. The two Zukertort Chigorin games went in opposite directions: a long win by checkmate on move 67 against julienk8 (1187), with 88.9% accuracy, and a 24-move loss against Ha1N4m (1230) at 57.4%, where one rook move on move 18 swung the evaluation from +4.18 to −5.23 and I was checkmated six moves later.
Game 2066: the knight that wasn't pinned
With White against Vansh13579 (1226), in a Queen's Gambit Declined, I was winning on move 11 (+3.23). Black had castled on the queen's side, and my bishop could attack the knight on c6.
Bishop to b5 attacks the knight, but the knight isn't stuck. It jumped to d4, taking my pawn, and from there it attacked my queen on b3, which had no defender, plus my knight and my bishop. I had to trade knights and the advantage was gone: the evaluation fell to −0.03, about equal. Rook to c1 puts pressure on the same knight without giving it a free square with a capture. Earlier in the same game I had also missed taking a pawn on b7 with my queen while I was up +7.13. I lost on move 27 (game 2066).
Game 2067: recapturing with the wrong piece
With Black against zakibn12 (1195), the position on move 13 was about equal (−0.47). White's bishop had just taken my knight on f6, and I had two ways to take back.
Taking with the bishop also lines it up on White's rook in the corner. But my knight on e4 was attacked by White's knight on d2, and it stayed attacked. White took it, I took the rook, White's queen took my bishop. The evaluation after my recapture was +2.77 for White. Taking back with the knight solves both problems at once: it recaptures, and it moves the attacked piece out of danger. I still won, on move 52, after three more blunders of mine and five of my opponent's (game 2067).
Game 2068: the free pawn
With White against berf4 (1202), in an Indian Game, I had castled on the queen's side and was clearly better (+2.78). Black's b-pawn on b5 was there for the taking.
I took it, and the bishop also attacked Black's rook. The engine still disagreed: the evaluation dropped to +0.42, and its best reply for Black was a bishop move that attacks my bishop on b5, which had no defender. Pawn to g4 instead starts pushing pawns toward Black's king. In the game Black moved the rook and I won on move 32 (game 2068).
The lesson of the week
After any capture, the position changes for both sides, and that's exactly when I stop looking. My checklist for captures is now two questions. First, which of my pieces is attacked right now? If one is, I try to capture with that piece (game 2067). Second, after I capture, what can their attacked piece do? A knight that is attacked but not pinned will take something on its way out (game 2066).
The other losses were the Pirc Austrian Attack against habqueen (1227), where a knight move on move 18 took the evaluation from +0.60 to +5.19, and a London System against mwehd (1215) in which I played 86.7% and my opponent 94.2%.
All games this week
| Game # | Date | Colour | Opponent (rating) | Result | Opening | Accuracy | Link |
|---|---|---|---|---|---|---|---|
| 2066 | Apr 27 | White | Vansh13579 (1226) | Loss | Queen's Gambit Declined | 64.8% | game |
| 2067 | Apr 28 | Black | zakibn12 (1195) | Win | Queen's Pawn Opening | 84.1% | game |
| 2068 | Apr 29 | White | berf4 (1202) | Win | Indian Game | 86.2% | game |
| 2069 | Apr 29 | Black | habqueen (1227) | Loss | Pirc Defense, Austrian Attack, Weiss Variation | 75.0% | game |
| 2070 | Apr 29 | White | julienk8 (1187) | Win | Queen's Pawn Opening, Zukertort Chigorin Variation | 88.9% | game |
| 2071 | Apr 30 | White | mwehd (1215) | Loss | London System | 86.7% | game |
| 2072 | May 1 | Black | tictacgburg (1196) | Win | Modern Defense with 1 e4 | 81.7% | game |
| 2073 | May 3 | White | Ha1N4m (1230) | Loss | Queen's Pawn Opening, Zukertort Chigorin Variation | 57.4% | game |
← Week 115 · Week 117 → · Back to the journey hub. Want the same turning points from your own games? Here's how the analysis works.
Frequently asked questions
Why do I keep blundering pieces?
In my week 116, every key mistake happened around a capture. In game 2067 I took back with the wrong piece and left my other knight hanging; in game 2066 my bishop attacked a knight that could jump away with a capture and an attack on my queen.
Which piece should you recapture with?
Look for the piece that is already under attack. In game 2067 my knight on e4 was attacked, and recapturing with that knight kept the game about equal; recapturing with the bishop let White take the knight.
Is it bad to grab a free pawn?
Not always, but it costs a move. In game 2068 I took a pawn on b5 and the engine's evaluation fell from +2.78 to +0.42; pushing my g-pawn toward Black's king kept a clear advantage.