4
votes
2answers
135 views

Machine learning for non optimal behaviour

I was working on the pricing of complex bermudean swaption when I noticed that the exercise is often (very) subobptimal. It seems that the clients are more sensitive to past growth or drop in rates ...