[pesenti]

My bet is that your model is no better than (weighted) chance to predict the next game. Just run the model on the years before 2012 and you'll see what I mean.

[Dwolb]

On my phone right now so I can't put this analysis on paper, but this feels a lot like looking at the associative property of addition.

i.e. the author regrouped the data to come up with the same result