toan6ca vs t3gantry

for ♝ James Claude Opus 4.8 2026-07-22

Five seconds on 12...h6 gave back most of a winning advantage, and the move right after it you found the best move on the board and never let go again.

The doubled c-pawns after 10.Bxc6 bxc6 bought you the b-file and a solid d5, and by move 12 the engine had you better by a pawn and a half. Then one quiet move nearly undid it.

12...h6 — the poke that invited the wrong trade

Five seconds, and the position went from -1.6 to -0.3. Poking the g5 bishop looks free, but it asks a question White answers well: Bxe7 Qxe7 Na4 Rab8 b3 Rfc8, and the pin resolves on his terms with your knight off the board.

f6 asks the same question and builds while it asks — f6 Bd2 e5 Ne2 e4 Qc3, still -1.5, with a pawn centre rolling and the f3 queen running short of squares. One move only annoys; the other annoys and gains.

Before a pawn move that touches one of his pieces, name the trade it invites and ask whether that trade is one you want.

13...Qb6 — two targets, and a bishop with no square

13.Bh4 kept the bishop on, and Qb6 was the engine's first choice against it: b2 and d4 attacked at once, and nothing good for the bishop to do. The line ran Bg3 Bxg3 Qxg3 Nf5 Qd3 Qxd4 and the position went to -2.2 — the whole advantage back, plus interest, in one move.

This is the scan working. You looked at what his piece could and couldn't do before you chose, which is exactly what didn't happen one move earlier.

15...Qxd4 — good enough, and there was better

After 15.Na4 the knight had walked away from its defenders. The engine's answer was Qa5 Bf6 Qxa4 b3 Qb5 c4 — you take the knight and ride out the pawn hits a piece up. You played 15...Qxd4 instead, which won the d4 pawn, took the queens off, and led to an ending you converted without a single flagged move. Nothing wrong with it; it just wasn't the most.

When he moves a piece to the edge of the board, count its defenders before you count anything else.

Phase notes

The conversion is the part of this game worth keeping. 18.Bg3 Nxg3 19.fxg3 handed you 19...Bxg3 — a second pawn and his kingside cover in the same move — and from move 15 to the end the engine flagged nothing you played. Rooks to the open files with 20...Rfc8 and 21...Rab8, then the break with 23...c5 and 24...c4. That is what technique looks like, and it used to be the missing part.

Profile check

Is the quiet-position scan still leaking? Yes, once. 12...h6 was five seconds in a position with nothing forcing in it, which is the same shape as the miss from the last pair: the scan fires when the board looks tactical and sleeps when it looks calm. What has changed is the other half of the file — converting a won position, which used to be the frontier, has now held twice running.

What to drill next

Quiet-move puzzles — positions where the best move isn't a check or a capture. That is where both of your recent errors have come from.
The position before 12...h6 — find f6 and play out f6 Bd2 e5 Ne2 e4 Qc3. Beat the engine from here three times running.
Hanging-piece puzzles — the Qa5 shape: a piece that has stepped away from everything defending it.

When nothing is forcing, spend the ten seconds anyway and say his best reply out loud before you commit.

for ♞ David Claude Opus 4.8 2026-07-22

In six moves the engine's first choice was a dark-squared bishop move four times, and each time you played something else — by the time the bishop was finally traded, it was on Black's terms and it cost a pawn.

Nothing you played before move 10 was flagged. The game was decided by one recurring question you kept answering with a different move.

10.Bxc6 — the wrong bishop

Forty-three seconds, and the position went from -0.9 to -1.6. Bxc6 gives up the bishop that had a job for a knight that was about to be traded anyway, and the recapture bxc6 hands Black the open b-file and a pawn chain that makes d5 permanent. The line ran Qd3 Bxf3 Qxf3 Qb6 Qd3.

The engine wanted the other trade: Bxe7 Nxe7 Bd3 Nc6 Nb5 Be7 — the g5 bishop for the e7 knight, the light-squared bishop dropping back to d3 where it has a diagonal, and b5 poking at c7 and d6.

The test isn't which capture is available. It's which of your minor pieces has the worse next ten moves. With Black's pawns on d5 and e6, that was the dark-squared bishop, and it was the one that stayed.

13.Bh4 — the same trade, now offered to you

Thirteen seconds, -0.3 to -2.2, and the biggest swing in the game. James's 12...h6 had just given the position back to nearly level and asked the bishop a question. Bxe7 Qxe7 Na4 Rab8 c3 Qf6 answers it and holds the balance. Bh4 keeps a bishop that now has one diagonal and no targets on it, and what it allowed was Qb6 Bg3 Bxg3 Qxg3 Nf5 Qd3.

The next move sharpens the point. 14.Qd3 took eighty-seven seconds — the longest think of the game — and the engine's move there was Bg3: same bishop, same question, third time of asking.

15.Na4 — the knight follows the bishop offside

Fifteen seconds, and -2.8 became -4.6. Na4 puts a knight on a square Black's queen is already looking at, and what it allowed was Qa5 Bf6 Qxa4 b3 Qb5 c4. Bg3 was still the move — Bg3 Nxg3 fxg3 Qxb2 Ne2 Qb5, taking the trade and accepting that b2 goes with it.

You did play Bg3 eventually, at move 18. By then the recapture fxg3 was forced and 19...Bxg3 picked up the pawn: the same trade, three moves later, on the worst available terms.

Phase notes

Opening: nothing flagged, and the position was still around -0.9 by move 10. In an Exchange Caro-Kann as White that is already the wrong side of level, which is worth a look on its own.

Endgame: 24.a3 in an already-lost position, forty-six seconds, -3.7 to -5.3. It gave the bishop something to bite on — c4 bxc4 Rxc4 Kf2 Bxa3 Rb1. When you're worse, a pawn move on the side where his bishop lives is a hook, not a defence.

Profile check

A new item, and it is the whole game: a minor piece with no good square is a piece to trade, not a piece to reposition. That is separate from the reply-check habit already in your file, though that one showed up too — 15.Na4 stepped onto a square the black queen was covering. The old pattern of long thinks producing the loose moves also held: forty-three seconds on 10.Bxc6, eighty-seven on 14.Qd3.

What to drill next

The position before 13.Bh4 — find Bxe7 and play out Bxe7 Qxe7 Na4 Rab8 c3 Qf6. Beat the engine from here three times running.
Caro-Kann theory and explorer — the Exchange gave you nothing here, and no single move was to blame for that.
Hanging-piece puzzles — for the 15.Na4 shape.

Before you reposition a minor piece, ask whether the position is offering to trade it — and if it is, take the trade the first time it's offered.

toan6ca 1655 vs t3gantry 1654 0-1

Analysis

♛ Deep reviews

12...h6 — the poke that invited the wrong trade

13...Qb6 — two targets, and a bishop with no square

15...Qxd4 — good enough, and there was better

Phase notes

Profile check

What to drill next

10.Bxc6 — the wrong bishop

13.Bh4 — the same trade, now offered to you

15.Na4 — the knight follows the bishop offside

Phase notes

Profile check

What to drill next

Key moments

Questions on this game