Matlo offers chess puzzles sorted by time control, blitz, rapid or long, each with its two-phase clock and its rating. Series start from your mistakes, missed puzzles come back with spaced repetition, and positions from your games become exercises.
A puzzle starts with thinking time, then a short window to play the line: in blitz, 30 seconds to think and 10 to play; in rapid, 1:30 and 15 seconds; in long, 5 minutes and 30 seconds. You calculate first, then execute, as at the board.
Each time control has its own Glicko-2 rating. A puzzle's rating stays the Lichess one: it acts as a reference opponent, and only yours moves after an attempt.
Puzzles come from the Lichess public database, under the CC0 licence, picked around your rating. A series covers a single motif: the one your profile points to, the one you just missed twice in the session, or the one you choose.
Puzzles from your games make up their own session: positions where you missed a winning move, and those where your opponent punished your mistake. After a wrong move, the correction names the motif and shows the opponent's best reply.
A missed puzzle comes back the next day, then later and later, until three solves in a row: that is spaced repetition, driven by the FSRS algorithm. “To redo” tells you how many puzzles are due, and the daily plan puts them before your session.
The daily plan chains, behind a single button, your due repertoire positions, a wasted advantage and your puzzle session, with a goal of 10, 20 or 50 exercises a day.
A position where you had at least half a point of winning chances ahead, lost without ever coming back, is a wasted advantage. You replay it against Stockfish, which defends its best, until you convert: a queen up, a mate, or the advantage held for ten moves outside the endgame.
Every wasted advantage comes back with spaced repetition and leaves the queue after three conversions in a row. On 257 test games, Matlo finds 46, 23 of them in the endgame.
Endgame puzzles from the Lichess database make up their own series: rook, pawn, queen, bishop and knight endgames, or all endgames. After a wasted advantage in the endgame, Matlo suggests the series of its family, in the game's time control.
Every puzzle's correction can be replayed on the analysis board, opened just before the mistake: the move played and the expected move as arrows, the solution as the main line, your wrong move as a variation. The engine can be turned on there to check another idea.
The puzzles come from the same public database, but Matlo picks the ones that help you: your motifs, your games, the ones you missed. The two-phase clock and the rating per time control are Matlo's own.
Puzzles keep their Lichess rating, so the scale is the same. Your rating, however, depends on the time control: you don't calculate the same way in 30 seconds and in 5 minutes.
No. Series, puzzles from your games and puzzles to redo are trained without a rating; only the free session moves it.
The correction menu lets you report it: another move also wins, the solution is wrong, or it is pointless. It won't come back to you.
Link Lichess or Chess.com: your games are analysed within minutes, then your first puzzle is waiting.