At least my initial analysis of the data seems to say that degrading instantly is better than any other time constant 😭. (May just be some nasty bug in the way we’re calculating probabilities?)

Reply to this note

Please Login to reply.

Discussion

Err, no, sorry, that’s wrong. Degrading instantly is bad (the learning does help!), but the model itself is worse than the naive “I dunno, 50/50 always” “model”, even when you learn.

(The LDK “historical model”, however, seems to do okay, keeping histograms of the liquidity bounds)