Should "Answer Buttons" and "True Retention" stats be this different?

TL;DR: Stats in “Answer Buttons” and “True Retention” sections are more different than I assumed after rescheduling at FSRS parameter optimisation. No other apparent issues with Anki or FSRS were observed.

Am I misunderstanding something regarding the true retention stats? When i previously picked “1 month” in the “Answer Buttons” section, then the Pass and Fail counts of Young and Mature cards matched the corresponding stats in the “True Retention” section. Yesterday I optimised FSRS parameters and rescheduled a deck with 3k reviews, which seemed to have worked correctly, where the average card stability went from 20 to 12 days (not an issue). However, now I noticed that the “Answer Buttons” and “True Retention” sections do not match for the last month. Specifically, shouldn’t the number of Pass and Fail presses for the last month for mature cards be quite similar between these two sections? This mismatch can be seen in the screenshot:

These stats were both in Anki 24.11 on Windows and in AnkiDroid 24.11.

Perhaps the “Answer Buttons” and “True Retention” sections consider differently previous reviews, where one way is affected by the rescheduling and the other is not. If this is the intended behaviour, then there may not be an issue, but otherwise perhaps there is a bug in the stats.

1 Like

No idea, I can’t repro either. @L.M.Sherlock Can you take a look?

I remembered now that I had saved the stats from just before the rescheduling. They showed close to the same stats as the image in the top post. Thus, the rescheduling did not have anything to do with the difference in stats, while the difference between stats sections is still present. Sorry for the confusing posting.

What are your learning and relearning steps?

Learning steps are “50s 10m” and relearning steps are “8m”.

True retention only considers the first review of the same cards in the same days.

I looked some more at the stats. I counted 11 mature reviews in the “Reviews” section for one month in the picture, which is the same number the “Answer Buttons” section reported for the total amount of Mature reviews.


In the picture the large amount (166) of filtered deck reviews 15 days ago were done with rescheduling and here all not-suspended cards were reviewed at once. Thus, it seems like the “Answer Buttons” section only reports reviews done in the selected deck, while the “True Retention” section reports all reviews done on cards in the deck, regardless if the review was done in a filtered deck or not.

1 Like

It’s the same for “Answer Button” too. Reviews in learn state is seperated from the stats for young/mature.

Screenshot-2024-12-15-23-08-59-05-a9eef3a2a561b80d5c76daebd0f9a14c

If the main difference between the “Answer Buttons” and “True Retention” sections is here the ex-/inclusion of reviews in filtered decks and there is a desire to combine these sections [ Let's make the True Retention table look better ], then I would prefer to just have the “True Retention” section behaviour, with filtered deck reviews included. Also, two sections with similar data causes confusion (at least for me). Just my two cents as an user. (:

1 Like

The True Retention table was just added to Anki with the most recent release and it is actively being redesigned as we speak – Let's make the True Retention table look better - #44 by rossgb . Once the dust settles on that, I’m sure thought will be given to whether Answer Buttons and True Retention need to continue to exist side-by-side.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.