What does their last move attack?
Your 109 finished blitz games, Oct 6–27, 2026 (19 Lichess, 90 chess.com, all 5+0), analysed by Stockfish 19.
Before you touch a piece, ask: “What did their last move attack?”
Check three things in two seconds: (1) the piece that just moved: what does it hit now? (2) the line it opened: what does the piece behind it see now? (3) a capture with check for you before you recapture.
If one of your pieces is attacked and not defended, saving it (or a bigger threat of your own) comes first. Your plan can wait one move.
The pattern
In blitz, your biggest single source of lost games is leaving a piece where it can be taken for free, in the middlegame. Usually it isn’t a careless new move. The opponent’s previous move attacked one of your pieces, and you played something else.
The most common shape is simple. Your opponent makes a move that looks like it is doing its own thing: a knight jump, a pawn push, a capture. But it also opens a line or lands a piece on an attacking square, and your queen or a minor piece is now hit. You continue with your own plan, and the piece goes.
The numbers
- 308 of your moves were “blunder-sized”: they dropped your winning chances by at least 0.30 on Lichess’s −1…+1 scale, which is about 15 percentage points of win probability. They occurred in 101 of the 109 games.
- The biggest group, by total winning chances thrown away, is hanging piece × middlegame: 42 moves across 34 different games, with a combined win-chance loss of 38.4 on that −1…+1 scale. It ranks #1 whether or not recent games are weighted more heavily, and it is well over the 5-occurrence bar for calling something a weakness.
- 26 of those 42 were cases where your piece was already under attack before your move, and your move neither moved it nor defended it.
- 9 of the 42 were your queen.
- Across all phases, hanging-piece moves number 72 in 51 games, and 37 of them were an attacked piece left where it was.
- The clock was not the cause. Only 3 of the 42 middlegame cases happened with under 30 seconds left. You had plenty of time; you just didn’t look.
- Where the games went: you lost 65 games in the window, and 34 of those 65 contained at least one blunder-sized hung piece.
Your puzzle and tactics ratings sit well above your blitz ratings: you can find tactics when you know one is there. The leak happens when nobody tells you to look.
Drills
Exact positions from your games. You are the side to move. One question solves each: what is attacked? Tap a piece, then a square (or drag).
Review
Six cards. A miss brings a card back tomorrow and restarts it. A correct answer moves it up: 3 days → 7 days → 21 days, then it retires. First review: Oct 28, 2026. Progress is saved in this browser only.
Your games
Every finished game in the window, Oct 6–27, 2026.
| Platform | Games | Time control | Result (you) | Rating in window |
|---|---|---|---|---|
Lichess adam5275 | 19 | 5+0 blitz | 8 W, 11 L, 0 D | 1298 → 1056 (before each game); 1078 after the last |
chess.com aglucky | 90 | 5+0 blitz | 34 W, 54 L, 2 D | 800 → 587 (after each game; 587 = live profile) |
| Total | 109 | 42 W, 65 L, 2 D |
All 109 games fetched were analysed.
- Lichess: blitz 1078 (555 games, RD 81, recent progress −17); rapid 1144 (provisional); puzzles 1403 (provisional). Your blitz fell −220 in the window (1298 → 1078). Nearly all of it came from the first two games on Oct 13 (−150 and −89), after a gap since December 26, 2025, so your rating was uncertain and swung hard. Since then it has gone from 1059 to 1078, about +19. Read it mainly as a recalibration, not a slump.
- chess.com: blitz 587 (best ever 1235, Feb 2021; lifetime blitz record 52-95-3); rapid 836, last played 2025-06-29; tactics best 1685. Your blitz fell about −213 in the window, from 800 (after your first in-window game) to 587.
Other patterns spotted
Each count is blunder-sized moves (win-chance drop ≥ 0.30 on the −1…+1 scale) with that pattern as its main tag, across all 109 games. Anything under 5 is not called a weakness.
| Pattern | Moves | Games | Note |
|---|---|---|---|
| Hanging piece × endgame | 13 | 12 | 8 of 13 had under 30 s on the clock. This is the one place where time trouble clearly matters: in the endgame you hang pieces when short of time. |
| Hanging piece × opening | 17 | 15 | The same lesson, earlier in the game. |
| Undefended pawn left en prise × middlegame | 21 | 15 | A smaller version of the same habit. Undefended pawns overall: 75 moves in 48 games, counting every blunder-sized move that had one at all. |
| Missed capture × middlegame | 17 | 14 | The best move was to take something, and you didn’t. That is the mirror image of the main lesson: their piece was hanging. |
| Walked into a fork × middlegame | 14 | 11 | The opponent’s best reply hit two of your pieces (≥ minor piece, or king + piece). |
| Fork × opening | 10 | 10 | Same as the row above, in the opening. |
| New pin on your piece (middlegame) | 8 | 7 | |
| Allowed mate in 1 | 3 | 3 | Below 5: not a weakness. Two of the three were back-rank mates. |
| Missed mate in 1 | 1 | 1 | Below 5. |
| No motif detected × middlegame / opening / endgame | 58 / 54 / 12 | 40 / 43 / 7 | Real losses of winning chances that the simple pattern checks don’t name: mostly deeper tactics or positional errors. |
The clock is not your main problem. Only 17 of 308 blunder-sized moves happened with under 10% of your time left (30 s of 300). You did, however, lose 10 of your 65 losses on time.
Method
Games from the public Lichess and chess.com APIs, fetched 2026-10-27. Stockfish 19 at 200,000 nodes per position; puzzle positions re-checked at 1,000,000 nodes, MultiPV 3. Win chance uses Lichess’s formula. Motif tags and chess.com phase labels are heuristics and can be wrong on individual moves.
Board: chessground (GPL-3.0-or-later). Move rules: chess.js (BSD-2-Clause). No accounts, no tracking; review progress stays in this browser.