Personal Tactics

The Personal Tactics feature extracts puzzle positions from your own analysed games and queues them as spaced repetition flashcards. Because the positions come from your actual play, they reflect the exact types of mistakes you make.

Requirements

Tactical puzzles are sourced from blunders and mistakes flagged during batch engine analysis. You must run batch analysis on your games before any puzzles are available. See Automated Analysis.

Each time you batch-analyse more games, new puzzles are automatically added to your queue.

The lobby

Navigate to Tactics in the left sidebar. The page opens to a lobby — it does not start a session automatically. The lobby shows how many puzzles are due, your lifetime accuracy, and the total number of puzzles in your database, with three controls:

• Start session — begins drilling (disabled when nothing is due).
• History — opens the drill history (see below).
• A settings (gear) button — opens session options:
  • Strictness — how close to the engine’s best move your answer must be: Strict (best move only), Standard (within ~30 centipawns, the default), or Lenient (within ~100 centipawns).
  • Solution linger — how long the solution is shown before auto-advancing (or set to Manual).
  • Show puzzles for — include Blunders, Mistakes, and/or Inaccuracies (blunders and mistakes on by default).
  • Exclude “from bad to worse” — hide puzzles from positions where you were already losing, so you drill the mistakes that actually cost you.

During a session

The board shows a position from one of your games — the move before your mistake, oriented to the side you played. A context strip shows the classification, the move you originally played, and the source game (“vs. {opponent} · {date}”) with a View game link that opens the full game on the Home page. Your task is to find the move the engine recommends.

1. Make your move on the board.
2. If correct: a green strip confirms it, shows the evaluation swing, and displays the full solution line. The next puzzle loads after the linger delay (or when you click Continue in Manual mode).
3. If incorrect: a red strip names the best move and draws a green arrow for it. Play that move to continue.

Sessions are not capped at a fixed number of puzzles — they continue until nothing is due, fetching more in batches as you go. The titlebar shows a running Reviewed and Due count, and an End button finishes the session early and jumps to the summary.

Session summary

When the session ends (all due puzzles reviewed, or you press End), the summary screen shows how many puzzles you reviewed and your percentage correct. Click Done to return to the lobby.

Drill history

The History view (from the lobby) is a paginated log of every puzzle you have reviewed: the date, the opponent, the classification, whether you got it right, and a View game link to open the source game. Use Load more to page through older entries.

Spaced repetition scheduling

Puzzles use the same FSRS v4 scheduler as the opening drills. Each puzzle position has its own card with tracked stability and difficulty.

• Correct on first attempt (Good): interval grows; the position is shown less frequently as you demonstrate recall.
• Incorrect (Again): the card lapses; the position re-enters the learning queue and is shown again sooner.

Puzzles you consistently miss will appear in nearly every session until your recall stabilises. Puzzles you reliably solve are spaced out to longer intervals automatically.