Portuguese Top League · 2026.07.17 · Elo 2400/2593 · defensive · ★★★★
The daily position for Saturday 12 September 2026. It is still open to solve — the grader reads your whole line, sidelines and evaluation included, before anything here is shown. Cover the solution and solve it first →
White to move. the model gives this to 1 in 6 at this level.
The solution
65.♖︎f8!(not65.♖︎c8?♔︎e466.♖︎xc5♖︎f667.♖︎c8♔︎xe3−+)65…♔︎d6(if65…♔︎e4then66.♔︎xc5=)66.♖︎xf5♖︎e4+(66…♖︎xe3is as good:67.♖︎d5+=)67.♔︎c3♖︎xe3+(67…c4is more resilient:68.♖︎f8♖︎xe3+69.♔︎xc4=)68.♔︎c4♖︎e4+69.♔︎d3=
65.♖︎f8 — almost nobody finds this
66.♖︎xf5 — the standard move
67.♔︎c3 — almost nobody finds this · worth +1
68.♔︎c4 — the standard move
69.♔︎d3 — hard to find · worth +1
“hard to find”, “the natural move” and the rest describe how often this move is played by Maia-2, weighted across the 1500/1900/2000 audience the difficulty model is fitted on — a model of human play, not a count of anyone who solved this position.
Members can take this away as a PGN — the diagram, the verified line and their own analysis in one file, for a database, Anki or Chessable.
What happened at the board
Galego played ♖︎h8 instead · 0-1
in the game65.♖︎h8♔︎e466.♖︎h3♖︎d667.♖︎h80-1full game ↗
The field
29% of the 14 committed lines here found the key move — the model predicted 16% at this level.
Some of the software based analysis is very strange in 65...Kd6 66Rxf5 Re4ch line it is not clear why giving up pawn with kc3 Rxpch is better than kd3, and then after giving up pawn playing kd3 is better than kc3