The story of the game
David, you played 92.5% accuracy over 75 moves against the family grinder and lost anyway — so let's be precise about where. Dead equal through 17.Qe2, only a shade worse by 19.Qf3, and never busted until move 52. The whole result came down to one idea the engine begged for twice and never got: push the c-pawn. This loss was one resource away from a draw.
The key moments
28.f4 — right idea, wrong executor (48% → 36%). Forcing the queen trade to kill the pressure was correct. But 28.Bf4 does the same job with a piece: after the trade the bishop stands active and the structure stays clean. After 28.f4 the trade still came (28...Qxd3), but that pawn on f4 left dark-square holes the black knight lived in for the next twenty moves. When two moves achieve the same goal, prefer the one that improves a piece.
51.Rc2 and 52.Bd4+ — the c-pawn was the draw (41% → 12%). Twice in two moves the engine wanted 51.c4 — your own passed pawn, your own counterplay, enough activity to hold. Instead 51.Rc2 defended passively, and 52.Bd4+ lost material by force: 52...Nxd4 53.cxd4 Rxd4 collected the pawn, and the even sharper refutation was 52...Rxd4 followed by ...Ne3+, forking king and rook. Look at the shape of that tactic — a knight fork, the same #1 flag that runs through your games as Black. When you're worse, the first candidate is "what's my fastest counterplay?", never "what defends?"
70.Ka4 — the last drop (8% → 0). Even here 70.Kc5, running the king at the kingside pawns, keeps swindle chances against an imperfect conversion. In queen-versus-pawns endings the king runs; it doesn't shuffle on the edge.
What went right — quite a lot
- The opening plan was a clean success: 8.Bb5+ into 10.Bxc6+ gave you a safe, simple structure against the Caro-Kann, and at 17.Qe2 the engine read the position as exactly 0.00. Nothing to fix there.
- 46.Rxd5 — the engine has no complaint with the simplifying grab; your calculation in liquidations is sound.
- One blunder in 75 moves. Your floor is remarkably high; the "precision player" line in your profile comes from games like this.
Profile check
Confirmed twice over: the fork blind spot showed up on the other side of the board this time (52.Bd4+ walking into the ...Ne3+ family), and the passive-defense lean — three of your four costly moves were retreats in positions where an active pawn push held. Add to the strengths column: near-blunder-free positional play over huge move counts.
What to drill next
- "Counterplay first" checklist when defending: before any retreat, list every pawn push and every active rook move. 51.c4 draws this game.
- The fork scan applies to your own pieces too: before dropping a bishop or rook onto a square, check the knight circuits around it.
- King-activity endgame drills: a set of "run the king at the pawns" positions — until 70.Kc5 is reflex.