Backlog: Low Retention Rate for Mature Cards

[Keks]

Comparing anecdotal actual retention to FSRS average predicted retention

I have this 10000 cards-sized deck. I made a custom session to review all of them at once. Before I did my average predicted retention of it was 98.2% (desired retention was set at 95%) When I went down to reviewing, my ACTUAL retention was 94% after doing the math. So is the size of this discrepancy normal, and what are the error ranges of the average predicted retention should I take it into account the next time I read one? Note: The reviewing was carried out on an extended period of time …

You have already asked similar questions and received answers to them.