Why is modeling policyholder behavior so damn hard, and what can we do about it? A bit of background on why I’m asking this. I recently had the privil…