The keyword "new" in the context of GM preparation is vital. Relying on the same tactical puzzles found in apps (which are often recycled from games played 50 years ago) leads to memorization rather than improvement.
Here is how Grandmasters generate new calculation material:
Below are three compact PGNs with brief annotations you can load into a GUI and use to practice calculation. For each: try to calculate key variations before using engine.
Annotation: calculate after 8...Nxf2 — evaluate multiple captures, discovered checks, and resulting material. grandmaster preparation calculation pgn new
Annotation: calculate sequence from 16...Bxg5 — check sacrifices and mating threats.
Annotation: practice calculating pawn race and piece placement leading to simplified endgame.
A GM asks:
Calculate variations to a quiet position — not just “I win material.” GMs look for stabilized advantage.
Depth: Not 10 moves deep in all lines, but 5–6 moves with candidate move pruning.
Calculation is often associated with flashy middlegame tactics, but GMs know that the hardest calculation occurs in endgame transitions. Creating PGNs of complex endgame conversions—where technique meets calculation—is a trendy new focus. These are positions where there are no immediate checkmates, but a sequence of 10-15 precise moves is required to convert an advantage. The keyword "new" in the context of GM preparation is vital
Before a GM calculates, they list no more than 2-4 reasonable moves. Searching every legal move is impossible. The art is pruning—ignoring the bad so you can deeply analyze the promising.
The phrase "PGN" in your request highlights a modern shift in chess study. While the physical book is excellent, training with the PGN files of the exercises offers distinct advantages: