How Often Should You Review Chess Openings?

The short answer

Review a new line about four hours after you learn it, again the next day, and then at gaps that keep widening: three days, eight days, three weeks. Cap it at a month so nothing disappears entirely.

That is the schedule. The rest of this explains why each part of it is shaped that way, because the reasoning is what lets you adapt it when your week falls apart.

Why the first two reviews are close together

Forgetting is not steady. It is brutal in the first day and gentle afterwards.

A line you learned this evening is at its most fragile tonight. Leave it a week and you are not reviewing it, you are learning it again. So the first two intervals are short and fixed: a few hours, then a day. Two correct answers in a row is not yet evidence that you know something, so nothing is assumed until you have given them.

After that the gaps multiply rather than add. Each success is evidence that the line is holding, so the next test can afford to wait longer, and the multiplier itself grows slightly each time. A position you have answered correctly six times is genuinely secure and does not need your Tuesday evening.

What a miss costs you

More than most people expect, and that is deliberate.

A missed position goes back to a one-day interval. It also carries a penalty forward: the multiplier that decides all its future gaps drops, so even once you start getting it right again, the position climbs back more slowly than it did the first time.

This is the correct behaviour. A line you have failed once is a line you are likely to fail again, and treating it like a fresh success would put it back to sleep exactly when it needs attention.

The practical consequence is that honesty costs you nothing and cheating costs you a lot. Marking a position correct because you nearly had it teaches the schedule that you know something you do not, and you will meet the same position over the board with the same gap in it.

Deep moves need more attention

This is the part almost every review schedule ignores.

A move on the third ply of a line is reached in every game that opening occurs. A move on the fifteenth ply is reached only if the fourteen moves before it all happened, which is much rarer, and it sits at the end of a longer chain of things to remember.

So depth should shorten the interval. A move deep in a variation is worth reviewing sooner than a move near the root at the same level of confidence, because there is more standing between you and it.

A rough version you can apply by hand: reduce the gap by about five per cent for every ply of depth. At ply fifteen that works out at an interval roughly a third shorter than the same position would get near the start of the line.

Where the ceiling should be

Left alone, multiplying intervals run away. Six correct answers and the schedule would like to leave a position for two months, then four.

Cap it at about thirty days. Not because the memory would fail sooner, but because a repertoire is not a static thing. Openings shift, your own preference shifts, and a position you have not seen since spring is one you can no longer confidently say belongs in your repertoire at all. The cap is as much about keeping the repertoire under review as the memory.

Doing this without an app

It is entirely possible on paper, and worth doing at least once so you understand the machine.

Keep a sheet with three columns: the position, the date you last got it right, and the current gap in days. Sort by due date. Each session, take the overdue rows from the top. Correct answers multiply the gap by two and a half. Wrong answers set it back to one day.

You will find two things. The system works. And it becomes unmanageable somewhere around a hundred and fifty positions, which is roughly where doing it by hand stops being sensible.

Doing this with one

Anything that schedules spaced repetition will do this for you. If you want one built for opening trees specifically, Chess Prep Pro uses the intervals above, applies the depth penalty automatically, and publishes the exact constants it uses.

Once the scheduling is handled, the question stops being when to review and becomes how much there is to review, which is a different problem. We cover that in how big a repertoire should actually be.