Coach definition, not a live verdict. These pages show how the X-ray Coach defines this concept and where its thresholds sit, sourced from the coach's own code. Unlike the tactic pages, nothing here runs a live detector — example boards illustrate the rule, they don't execute it.

What it means

The exact rule the X-ray Coach applies

  • Eight moves, close together: the engine's top 8 moves must sit within 1.00 of each other. Fewer than eight playable moves — or a forced mate anywhere in them — means the position isn't quiet, so it doesn't count.
strategyGate.ts (withinLadder, MAX_LADDER_GAP_CP/LADDER_DEPTH)
  • Position not already decided: the evaluation before the move must be within ±2.00. Once someone is winning outright, the choice stops being strategic.
strategyGate.ts (withinEvalBound, MAX_POSITION_EVAL_CP)
  • Material within one pawn, counted as P1 N3 B3 R5 Q9. A wider gap is a different kind of decision.
strategyGate.ts (withinMaterial, MAX_MATERIAL_PAWNS)
  • No opposite castling: excluded when one king stands on the a/b/c files and the other on f/g/h. A king still on d/e belongs to neither wing, so it is not “opposite”.
strategyGate.ts (isOppositeCastling)
  • Middlegame only, and no tactic: openings and endgames are excluded by game phase; every remaining position is classified by Koga and anything tagged as a tactic is dropped. A quiet preparatory move is not a tactic.
strategyGate.ts (tacticVerdict) · quietPositionCore.ts (isEligibleMiddlegame)

How it's scored

Why it matters

David curated 65 positional middlegame positions as the reference set; this definition admits 64 of them. The one it turns away is Botvinnik-Yudovich (URS-ch08) at +2.05 — five hundredths outside the ±2.00 line of the definition itself. The rule it replaced admitted only 59 of the 65. On real student games the gate keeps roughly one middlegame position in six.

Related