Anki vs FSRS for the SIE: Which Spaced-Repetition Algorithm Wins?

What is spaced repetition, and why does it matter for the SIE?

Spaced repetition is a study technique where flashcards reappear at increasing intervals based on how well you remember them. Get a card right easily, you see it again in two weeks. Get it wrong, you see it again tomorrow. Over time, the algorithm learns which cards need more practice and which can fade into the background.

For the SIE, this matters because the exam covers a wide vocabulary surface: regulators, products, account types, prohibited activities, customer protection rules, and dozens of definitions that all sound vaguely similar until you’ve drilled them. Reading through a textbook three times will not lock those terms in. Spaced repetition will.

The question is: which spaced-repetition algorithm should you use?

What is the SM-2 algorithm Anki uses by default?

SM-2 is the algorithm Piotr Wozniak published in 1988 for SuperMemo. Anki adopted a modified version of it as its default scheduler.

The basic mechanics:

• Each card has an “ease factor” (how easy you find it).
• When you rate a card “Good,” the next interval is the previous interval multiplied by the ease factor.
• Rate it “Hard” or “Again” and the ease factor drops; rate it “Easy” and it rises.

SM-2 works. Tens of millions of medical students, language learners, and exam takers have used it successfully. But it has known weaknesses:

• It treats every card the same way, regardless of how consistently difficult that card is.
• The “ease factor” only goes one direction at a time, so it can drift into “ease hell” where mature cards keep ballooning intervals even though you actually keep forgetting them.
• It schedules more reviews than necessary at equal retention rates.

In other words, it’s a 36-year-old algorithm built on guesses about how memory works, not on data.

What is FSRS v5?

FSRS stands for Free Spaced Repetition Scheduler. It’s a modern algorithm developed by Jarrett Ye and refined through open research. FSRS v5 (released 2024) is the current version.

The key differences from SM-2:

• FSRS uses a three-component model of memory: stability (how durable a memory is), difficulty (how hard a card is for you), and retrievability (probability you’ll recall it right now).
• It’s trained on data: parameters are fit to your actual review history, so the model adapts to your memory, not an average user’s.
• You set a target retention rate (e.g., 90%) and the algorithm schedules reviews to hit that target with minimum work.

In published comparisons on Anki review logs, FSRS produces 15 to 40% fewer reviews at equal retention compared to SM-2. For a timed study window like SIE prep, that translates directly to less time spent reviewing and more time spent on practice questions.

Which algorithm is better for the SIE specifically?

FSRS, for three reasons:

1. The SIE has a hard deadline. Most candidates study for 2 to 6 weeks. SM-2’s calibration assumes you have months or years to let intervals stretch. FSRS adapts to a compressed window better because you can set a higher target retention (say, 95%) for the weeks leading up to test day.

2. SIE content is dense and similar. Many SIE concepts sound alike: Reg T vs Reg U, Rule 144 vs Rule 144A, MSRB vs FINRA jurisdiction. Cards in this kind of dense semantic field tend to “leech” under SM-2 (you keep forgetting them but the interval keeps growing). FSRS handles leeches more aggressively because difficulty is part of the model.

3. The exam is timed. SIE practice exams reward fast recall. You want flashcards scheduled at the moment recall is just barely possible, which is exactly what FSRS targets. SM-2 schedules them later than that on average.

Can I just use FSRS inside Anki?

Yes. As of Anki 23.10 (October 2023), FSRS is built in as an optional scheduler. To switch:

1. Open Anki Desktop (FSRS is not yet on AnkiMobile or AnkiDroid for some versions, check your version).
2. Go to Tools → Preferences → Scheduling.
3. Enable “FSRS.”
4. For each deck, go to deck options and click “Optimize FSRS parameters” (this trains the model on your review history).
5. Set your target retention (90% is the default; raise to 92 to 95% as you approach test day).

This is a real upgrade, and the rest of this article applies whether you use FSRS via Anki or via a purpose-built tool.

What’s the catch with FSRS?

A few real ones:

Cold start. FSRS works best after about 1,000 reviews of training data. If you’re using Anki for the first time during your SIE prep, you’ll be on the default parameters for the first week or two. They’re still better than SM-2, but not as tuned as they’ll get.

You still have to make good cards. FSRS optimizes scheduling, not card quality. A poorly written card (long, two-concepts-in-one, ambiguous answer) will still be hard to learn no matter what algorithm shows it to you.

Anki’s UI was not designed for FSRS. The “Hard / Good / Easy” buttons map awkwardly onto the FSRS model. Some power users prefer pass/fail cards and a simpler grading scheme. Purpose-built FSRS tools are starting to ditch the four-button system entirely.

How much time does FSRS actually save?

The practical answer for an SIE candidate using a deck of around 600 to 800 cards over 5 weeks:

Numbers are illustrative, drawn from typical FSRS-vs-SM-2 comparisons in the published literature. Your mileage will vary based on card quality and your own grading consistency.

The headline: at the same retention level, FSRS shows you fewer cards. At a higher retention level, FSRS shows you about the same number as SM-2 would but with much better recall. Either way, you win.

Are there cases where SM-2 is fine?

Yes:

• Long-horizon study (a year or more, like medical school). The advantage of FSRS narrows as the time window grows.
• Very small decks (under 100 cards). Algorithm differences are washed out.
• You already have years of SM-2 review history on Anki and don’t want to retrain. (Though FSRS can use your existing history as training data, so this is a weak excuse.)

For the SIE? None of those apply. You have a 4-to-6-week window, several hundred cards, and probably no prior Anki history. FSRS is the right call.

What about other algorithms?

You may have heard of:

• SM-15, SM-17, SM-18: SuperMemo’s proprietary, closed-source algorithms. Used inside SuperMemo only. Probably better than SM-2 but you can’t use them outside that app, and SuperMemo’s UI is, charitably, a relic.
• Mnemosyne’s algorithm: A variant of SM-2. Similar performance.
• Custom intervals (Quizlet’s “Long-term learning”): Proprietary, undocumented. Most likely a simple SM-2 variant or a fixed-interval system. No published evidence it beats Anki.

FSRS is currently the strongest open algorithm with public benchmarks. That’s why Anki adopted it.

Should I switch mid-prep if I’m already using Anki SM-2?

Yes, especially if you’re more than a week out from your test date. Switching is non-destructive: FSRS reads your existing review history as training data, so you don’t lose progress. Worst case, you go from SM-2 efficiency to FSRS efficiency. There’s no downside.

If you’re within a few days of test day, the answer is murkier. FSRS will reschedule cards based on its model, which may bunch reviews differently than you expected. If you’re already on a workable Anki SM-2 schedule and your test is in 48 hours, finish on SM-2 and switch next time.

The bottom line

For SIE prep, FSRS v5 is meaningfully better than SM-2. You can use it inside Anki (free, just enable in preferences) or use a purpose-built tool that runs FSRS by default. Either way, expect 20 to 40% fewer reviews at equal or better recall than the Anki default.

The algorithm is only as good as the cards you feed it, though. A great FSRS scheduler with mediocre cards will still underperform a mediocre SM-2 scheduler with great cards. If you’re building your own deck, prioritize card quality first, algorithm second. If you’re using a curated deck, make sure the cards were written for the exam (not just scraped from a textbook).