@Thomas_Thumb wrote:A zero correlation would be scary - after all, a positive trend over time in one model usually translates to positive trends in other models.
Many factors are the same, just weighted differently. A lot is invested on the various flavors of credit scoring tools not being completely random. I would have been disheartened if correlation came in under 0.8. Good to know it's not all random guesswork.
Oh sure... I was expecting at least >0.5 - but if you had asked me an hour ago to guess, I would have gone with 0.7-ish.
Certainly not almost 0.9 - I'm honestly surprised it tracked as close as it did. (The absolute scores were fairly different, of course - averaging a 40 point delta.)