Entering MS2 in FSRS Review Hell (400+ reviews/day) – Need Optimization Advice

🙋 Support
1

Hey everyone,

Just finished MS1 using the AnKing Step Deck and I’m currently stuck in what feels like review hell. I have ~10,000 cards unsuspended so far. For the last few weeks post-exams, I’ve been averaging 400+ reviews a day, and it’s becoming unsustainable before MS2 starts.

My Current Habits & Setup:

  • Deck: AnKing Step Deck (FSRS on).

  • Grading: Honest. I press Again on misses, never Hard to try and save a card.

  • Leeches: Already have about 250+ leeches suspended.

I want to optimize my FSRS settings to make my daily load more efficient without tanking my retention for Step 1.

My FSRS Settings:

  • Desired Retention: 85%

  • Maximum Interval: 1825

  • My current true retention (from stats): 81.4%

What tweaks should I make to get this daily load down to a manageable number? I am concerned that my true retention is lower than my desired retention and that is why I am getting more reviews/day.

Thanks in advance!

2
Thinking

Search
Found 0 results for ‘tags:official FSRS review hell 400 reviews per day desired retention 85% AnKing Step Deck category::person_raising_hand: Support order:latest max_posts:10 status:open’

Search
Found 1 result for ‘tags:official FSRS desired retention 85% reviews per day order:latest max_posts:10 status:open’

You’re not doing anything “wrong” here—your situation is mostly a combo of:

  • Huge card load (10,000 unsuspended)
  • Honest grading (which is good!)
  • FSRS tuned a bit aggressively for Step-style studying

A few targeted changes can make this more manageable without wrecking Step 1 prep.

1. Adjust FSRS retention and max interval

For Step-style decks, lots of people find that 80–82% desired retention is a better balance than 85%, especially with big decks.

  • Change Desired Retention from 85% → 80–82%
    This will:

    • Lengthen intervals a bit faster.
    • Reduce the daily review load.
    • Still keep you in a safe range for exam prep (you’re already at ~81.4% true retention, so this aligns your settings with reality instead of fighting it).

  • Keep Maximum Interval high (1825 is fine)
    The max interval is not the main driver of review hell at your current stage. Shortening it would usually increase long-term load.

2. Limit what’s coming in, not just what’s already due

With 10k unsuspended and 400+ reviews/day, you should aggressively control new load:

  • New cards/day

    • If you’re currently doing > 50 new/day, consider dropping to 20–40 new/day until your daily reviews fall below a number you can tolerate consistently (e.g., 250–300).
    • It’s better to be slightly “behind” on unsuspending than to burn out and stop reviewing entirely.

  • Pausing non-critical material

    • In the AnKing Step deck, consider re-suspending lower-yield or very low-relevance tags (e.g., minutiae-heavy pharm or micro if your school/Step-style Qs don’t emphasize them as much yet).
    • Focus on the highest‑yield tags (AnKing hierarchies, Boards & Beyond/Sketchy/Pathoma cores, etc.).

3. Be strategic about leeches & lapses

You’re already suspending leeches, which is great. A few extra tweaks:

  • Don’t be afraid to delete/suspend “bad” cards
    If a card is vague, overly long, or keeps tripping you up despite multiple edits, it’s often better to:

    • Suspend it and replace it with a better version (or learn it from a resource now and revisit later), instead of grinding it repeatedly.

  • Edit confusing cards on the spot
    Even changing a few words to match how you think (mnemonic, simpler wording, bolding the key phrase) can drastically reduce repeats on those problem cards.

4. Tweak how you grade (without lying)

Your grading philosophy is solid, but for FSRS:

  • It’s okay to use “Hard” as “I barely knew this / needed hints / took a long time,” not only as a way to “save” a card.
  • If you’re extremely strict with “Again,” you can slightly overproduce short-interval reviews.
    • For things you almost knew, using “Hard” instead of “Again” can relieve some load while still not “rewarding” the card with a big interval.

You should still hit Again on true misses, of course.

5. Expect a lag before things calm down

Even with good settings, there’s an “after-exam bulge” of reviews:

  • You just finished MS1 with lots of new cards added over time → a lot of those are now hitting early and mid intervals.
  • Once you:
    • Lower desired retention to ~80–82%
    • Cut new cards/day
    • Tighten what you keep unsuspended

…your reviews should trend downward over 1–3 weeks, not overnight. What you’re feeling now is partly that lag.

If you’re okay sharing more

If you reply with:

  • Your average new cards/day over the last month
  • Peak daily reviews and what feels “sustainable” (e.g., 250? 300?)
  • Whether you’re using any filtered decks / custom study

I can give more concrete number-based suggestions (e.g., target new/day, what to re-suspend, and what retention to aim for in the months before MS2 + Step).

3

Hello,

I would be happy to help. :slight_smile:

There are only so many options when it comes to reducing review load, unfortunately, and Anki more or less follows the same rule/s as anything else under the Sun

You can do one, or a combination, of the following:

  • Introduce fewer new cards.
  • Suspend existing cards.
  • Adjust desired retention.

Those are about your levers, for better or worse. Pretty much everything will come down to some combination of those, in one way or another. Lowering your retention will, perhaps obviously, impact your retention (and so you’ll want to account for this when choosing a new value)

There isn’t a lot else to it, I’m sorry to say. But, please let me know if that answers your question, and if there’s anything else I can help with. :slight_smile:

4

It sounds like what you are saying is a normal experience? I was concerned that I had done something wrong. This summer, I haven’t had many new cards other than some I have been adding that are relevant to UWorld practice questions.

I am concerned that the system is pushing new cards out too far (15-20 days) upon the first “Good” designation—is this normal too? I feel like this may hurt the algorithm if I can’t remember a new card 2-3 weeks from now (which is likely)!

I can try lowering the retention rate. Do you think 82%, like the AI suggested, sounds like a good number? When doing this, should I select “Reschedule cards on change” or leave that alone. This setting seems daunting, like I could drastically mess up my settings or schedule.

Could you suggest a way to search for cards that may be the low-yield, less helpful type? If there is a certain tag that I should look for, that would be helpful.

5

400+ reviews per day before Step 1 is quite normal. My average per day before Step 1 was 600 per day, although this was with the old algorithm (SM-2). And nearing Step 1 it got to 1000 per day, which was unsustainable but I tried my best doing as much as I can. It made Step 1 very easy for me, so yes that number is normal.

What is your learning step? Or if you can take a screenshot of your settings that would be helpful as well. If the cards are too far out in the future when you first see them, then increasing desired retention will help. My FSRS currently is 88%, I experimented with 92% and then 90%, the sweet spot for me was 88%.

If you want to suspend low yield cards for Step 1, there is a Low/High yield tag in #AK_Step1_v12, go to low yield and relatively low yield subtag and suspend those. However, at this stage, I would encourage you to keep going with what you’re doing as long as its giving you good results, since 400-600 average is pretty normal depending on how many cards you unsuspend

6

I would not press that button, just keep as is and the cards will slowly adjust to the new desired retention overtime

7

Yeah, I would say that 400 review per day is well within the normal range. It’s not quite on the low-ish side, but it’s definitely not on the high end.

It’s hard to say that it’s abnormal. If your desired retention, from your description, is in the low 80’s that sounds entirely expected.

It won’t ‘hurt’ the algorithm so much as, if the algorithm sees over time that you’ve forgotten those cards it will adjust the intervals in the future. This is baked into how the algorithm functions, and so it will largely self-regulate in this way particularly over time

Generally it’s recommended not to tinker with that. You can see huge swings in number of cards due, and in the long run Anki will adjust to you eventually anyway. It can cause some headaches if you reschedule all at once, for a few reasons

It’s hard to say what is a ‘good’ number, honestly. At the end of the day the desired retention is just that: it’s the retention that you personally desire.

Obviously the higher that number is, the more stuff you’re likely to remember (though this comes at the cost of higher average daily review numbers). The lower that number is, the less work you’re likely to have on a given day but the more stuff you’ll forget. This is, for better or worse, an inherent trade-off ultimately

It’s probably fair to say that the vast majority of folks end up landing somewhere between 80-90%, on the whole. There are probably exceptions, I’m sure, but that’s likely to be most people most of the time

The yield tags, contained in the Step deck, are ideal for this:

CleanShot 2026-06-18 at 07.53.06@2x
CleanShot 2026-06-18 at 07.53.06@2x672×834 63.6 KB

You can use those to target, or suspend, cards based on relative yield. It won’t be a perfect heuristic, but it’s definitely handy for honing in on what’s important