100 75 50 25 4h d1 d3 d8 d22 d30 reviewed left alone recall %
Left alone, a line fades. Each review lifts it back and buys a longer gap before the next one.

How it works

  1. A new position is due straight away

    Anything you have never trained has no history to schedule from, so it goes to the front of the queue.

  2. The first two intervals are fixed

    Answer correctly once and the position returns in about four hours. Answer correctly again and it returns the next day. These two steps are the same for every position, because two correct answers is not yet evidence of anything.

  3. After that the gap multiplies

    From the third correct answer onward, the next gap is the previous gap multiplied by an ease factor. The ease factor starts at 2.5 and rises by 0.1 each time you get the position right, up to a ceiling of 3.5. So the gaps do not just grow, they grow faster the better you know the move.

  4. A miss resets the streak

    Get it wrong and the position is due again tomorrow, and the ease factor drops by 0.2, down to a floor of 1.3. A position you keep failing will keep coming back roughly daily until it sticks.

  5. Deep moves are pulled forward

    The interval is divided by a depth penalty of one plus five per cent per move of depth. A move fifteen plies into a line is reviewed noticeably more often than a move on the third ply, because there is more to forget on the way to it.

Under the hood The exact settings and constants. Not needed to use it.
Starting ease factor
2.5
Ease change per correct answer
+0.1, up to 3.5
Ease change per miss
-0.2, down to 1.3
First interval
About 4 hours
Second interval
1 day
Later intervals
Previous interval x ease factor
Depth penalty
Divided by 1 + (depth x 0.05)
Longest gap
30 days

Answer a shallow position correctly every time and the gaps run roughly 4 hours, 1 day, 3 days, 8 days, 22 days, then 30 days from there on. The cap exists so nothing in your repertoire disappears from view for a whole season.

Why this method and not another

The spacing effect

Hermann Ebbinghaus established in the 1880s that material reviewed at widening intervals is retained far better than material reviewed at a constant rate, or crammed. Everything above is an application of that one finding to a position tree.

The testing effect

Being asked to produce an answer strengthens memory more than reading the same answer again. This is why training asks for the move and waits, instead of playing it for you and asking whether it looked familiar.

What we do not claim

We have not run a study on Chess Prep Pro users, so there is no percentage here. The method is well evidenced in the psychology literature. How much it helps you depends on how often you actually open the app.

What it will not do

  • It schedules positions you have saved. It has no opinion on whether the moves in your repertoire are good ones, which is what the opening book and the engine are for.
  • It cannot tell the difference between a move you knew and a move you guessed correctly, so an honest session is worth more than a lucky one.
  • It needs regular short sessions. A month of nothing followed by an hour of catching up defeats the point of the schedule.

Try it on your own openings

Free to start, with no repertoire or move limit.

No account needed no repertoire or move limit · premium optional