Spaced repetition algorithms

A spaced repetition algorithm decides when to show you a card again after you review it. The goal is to schedule each review just before you would otherwise forget the answer, so you get maximum retention with minimum effort.

Different algorithms make that decision in different ways. Some use fixed boxes, some use fixed formulas, and some use machine-learning models of memory. Deckbase uses FSRS because it adapts to your personal recall history, but the other algorithms are still useful to understand — especially if you are migrating from another app.

Quick comparison

FSRS (Free Spaced Repetition Scheduler) is an open-source algorithm created by Jarrett Ye. It models memory for each card using three values — Difficulty, Stability, and Retrievability — and calculates intervals from your personal review history instead of applying fixed rules.

Because FSRS adapts to each learner, it typically reduces daily review volume by 20–30% compared with SM-2 for the same retention target. It is the default scheduler in Deckbase and is also available in Anki 23.10 and later.

• Adapts intervals to your actual recall patterns.
• Usually fewer reviews for the same retention.
• More complex than SM-2 or Leitner; benefits from more review history.

SM-2 is the SuperMemo-2 algorithm created by Piotr Wozniak in 1987. Anki used a modified version of SM-2 as its default scheduler for many years, and it is still available as a fallback in Anki today.

SM-2 tracks a single value called the ease factor. When you rate a card Easy, the ease factor increases and future intervals grow faster. When you rate it Again or Hard, the ease factor drops and intervals shrink. The algorithm is simple and robust, but it does not model memory directly and can push cards into "ease hell" if they are lapsed repeatedly.

• Simple to understand and implement.
• Works well for small to medium decks.
• Less efficient than FSRS for large, long-term libraries.

The Leitner system is the simplest spaced repetition method. Cards start in Box 1. If you answer correctly, the card moves to the next box and is reviewed less often. If you answer incorrectly, it moves back to Box 1.

Leitner is easy to explain and works without any software, which is why it appears in paper flashcard systems and basic apps. The downside is that intervals are fixed by box number, not by your actual recall probability. Two cards in the same box may be at very different memory strengths.

• Easy to learn and teach.
• Good for small, focused sets of facts.
• Not ideal for large decks or long-term retention goals.

Fixed interval scheduling shows cards at predetermined delays — for example, 1 day, 3 days, 7 days, 14 days, and 30 days after each successful review. This pattern is common in corporate training, language courses, and basic flashcard apps.

Fixed intervals are predictable and easy to plan around, but they ignore how well you actually remember each card. A card you find easy and a card you struggle with may both be scheduled for the same 14-day review.

• Predictable calendar.
• Easy to implement in simple apps.
• Wastes time on easy cards and may miss hard cards.

Why Deckbase uses FSRS

Deckbase chose FSRS because it is the only mainstream algorithm that adapts to your personal memory. SM-2, Leitner, and fixed intervals apply the same schedule to every learner. FSRS learns from your ratings and calculates intervals that fit your actual forgetting curve.

That adaptivity matters most when you study for months or years. Independent benchmarks on hundreds of millions of Anki reviews show FSRS reduces prediction error by roughly two-thirds compared with SM-2, which typically means 20–30% fewer reviews for the same retention.

Related resources