Bayesian model averaging: ranking of model weights and LOO don't match

[sammosummo]

This is perfect, thank you.

I guess I didn’t appreciate the difference in intentions of stacking and simple model comparison using LOO.