Structures, conversion, calculation
Diagnosis: You reach playable middlegames but can't convert small edges, your plans are vague ('attack something'), and calculation is inconsistent — brilliant one move, blind the next. The fixes: pawn-structure literacy, a conversion method, and a calculation routine.
1. Learn structures, not openings
8 minA dozen pawn structures cover most of chess: Carlsbad, IQP, hanging pawns, the Slav/Caro triangle, French/Advance chains, Najdorf-style e5-vs-e4, KID locked centre, fianchetto structures, symmetric open files. Each has known plans for both sides — known for a century.
When you know the structure, transpositions stop scaring you and unfamiliar openings become familiar middlegames. IQP positions play the same whether they came from the QGA, the Caro-Kann Panov or the Nimzo.
Study route: take one structure per week from the plan cards, play three drill games in it, then run the scan to see if your ACPL in that structure's games drops.
2. Converting +1: the boring wins
7 minUp a pawn or an exchange, players in this band attack harder — and lose. Conversion is the opposite skill: trade PIECES (not pawns), kill counterplay before taking anything else, improve your king, only then advance.
The technique list: rooks behind passed pawns; create a second weakness (one weakness is defensible, two overload the defence); 'do not hurry' — repeat once to reach the time control and to test their patience.
Biggest single leak: grabbing a second pawn while their pieces activate. Activity is worth more than a pawn at every level of chess — including yours.
Up a clean pawn. The winning plan is boring: secure the king, neutralise the active rook, only then push the b-pawn. Attackers lose this; accountants win it.
3. A calculation routine that holds under pressure
8 minCandidates first: list two or three moves BEFORE calculating any one of them deep. The one-line tunnel — calculating a single move ten moves deep, and playing it — is this band's signature error.
For each candidate: their most forcing replies first (checks, captures, threats against you). Calculate until the position is quiet, then evaluate: material, king safety, worst piece. Compare candidates at their quiet endpoints, not at their most exciting move.
Blundercheck the winner once — fresh eyes, one second: 'after this exact move, what takes what?' Then commit. Never re-open the analysis after you've stood up, metaphorically: flip-flopping burns clock and confidence.
4. Is the attack real? Count before you sac
6 minYou've learned attacks win games; now learn when they're actually there. Count attackers vs defenders in the king zone: you need more pieces IN the attack than they have defending, usually by two. Half-committed attacks rebound — every attacking pawn move weakens your own king for the endgame you just created.
Green lights: their knight left the kingside (an f6/f8 knight is worth two defenders), the centre is closed or you control it, you can open a file faster than they can trade queens.
Red lights: your own king airy, centre fluid, their queen eyeing a counterattack. With red lights on, the correct 'attack' is usually a queenside pawn break. The structure decides — again.
Two attackers pointing kingside, three defenders already home: this attack isn't real yet. Count bodies in the king zone before you commit a single pawn.