Remember if you get a bad model run:
1. When is sampling?
2. That model has been terrible this season.
3. It has bias in that direction.
4. It injected bad data.
5. Point out the one run in the ensembles that buries you.
6. Clearly an outlier. Be blind to the truth.
feel free to add