I found this article at Sporting News. To call this a prediction is (in a word) dishonest.

Regardless of the simulation being used, using a single example is like looking at one grain of sand on a beach. Other grains of sand may look the same, similar, or nothing at all like the individual grain being looked at.

While 'autonomous agent-based models' do have some value, they should be used methodically. For what it's worth, here is another humorous example (from the same author) of what the simulation demonstrates as a possibility.