I was thinking of adding something about random chance, but thought it would
be simpler to look first at what actually happened and add in the statistical details
afterwards. My first suspicion is that none of these "test predictions" would
pass the statistical muster.
I'd be interested to see how your program rates these.