Transparency

Training methodology

How adaptive selection, themes, Woodpecker cycles, spaced repetition, engines, and error-led analysis work together.

From diagnosis to practice

The recommended flow begins with positions near the player's estimated level. Results and observed difficulty influence later sessions. Themes are not decorative tags: they group recurring failures such as missed pins, loose pieces, and back-rank mates.

Adaptive rating

A puzzle rating estimates challenge; it is not an absolute measurement of a person. Selection seeks tasks hard enough to require calculation without turning the session into guessing. A small sample cannot support strong conclusions, so reports emphasize trends.

Woodpecker and spacing

Woodpecker training repeats a limited set in cycles, aiming for accuracy and faster recognition without losing understanding. Spaced repetition returns weak items after intervals. In both cases, the player should explain why the line works and why a tempting alternative fails.

Game analysis

Analysis starts without an engine: record candidates, plans, and moments of uncertainty. Use Stockfish afterward to test tactics and evaluation. Automated narrative is interpretive and can be wrong; compare it with the board and engine line. Suggestions are educational aids, not guarantees.

Limitations

What is automated and what is interpretive

It helps to separate two layers of the system. The automated layer handles what is objective and repeatable: selecting positions by rating and theme, scheduling reviews, validating that a move is legal, and checking a mating sequence with the engine. The interpretive layer handles what requires judgment: why a plan is good, which principle explains a mistake, what to study next. The first can be trusted to the machine; the second is a suggestion you should evaluate critically.

This distinction avoids two common extremes: dismissing the tool out of distrust, or accepting every engine number as absolute truth. The engine tells you which move is stronger; it does not tell you why you failed to find it or which habit to change. That "why" remains your work, and it is exactly where learning happens.

How to interpret AI and engine suggestions

Treat the engine's evaluation as a very strong second opinion on tactics and calculation, and a weaker one on long-term decisions, where small numeric differences rarely matter to a human player. When an automated narrative explains a position, compare each claim against the board and the engine's main line; if something does not hold up, trust what you can verify. Suggestions exist to generate good study questions, not to replace your reasoning or a coach.

None of these tools guarantees improvement. What produces progress is the consistent cycle of playing, reviewing honestly, training the weak theme, and returning to practice. ChessGrade's methodology organizes that cycle; the discipline to follow it stays with you.