Also,...
Over-fitting data.
My god these things drive me crazy.
If it’s real, then he should be able to answer this question posted on Twitter by Gregory Allen Perry ( @gallenperry ):
What is the payout on a $1000 bet that a terrorist attack in will occur in Kashmir in the month of February with no less than 20 victims killed if the price of silver drops by 5% immediately prior to Pakistani tax assessment in the year before?
If he can’t answer a question at least somewhat in the same vein as that, then this is crap. If he can (within any reasonable degree of confidence), then I’m wrong.