Remember this summer, how Cortana (the Microsoft personal assistant device) was perfect at predicting the knockout round of the World Cup? Well, in the immortal words of Seinfeld:
The app is going to try and predict every NFL game this season — if it was perfect, that would be insane — and here’s their logic:
For pro football, we model the respective strengths of the teams by examining outcomes from previous seasons including wins, losses, and the very rare tie outcome (two games since 2009), factoring in margin of victories, location of contest, playing surface and roof cover (or lack thereof), weather and temperature conditions, scoring by quarters, and multiple offensive and defensive statistics. In addition to this prior model, we identify fans on Web and Social sites and track their sentiment to understand the aggregate wisdom of this expressive crowd. This introduces data which statistics alone cannot capture, providing real-time adjustments which surprisingly can capture injury news and other substantive factors in win probabilities.
Bringing in web + social will make this a little more interesting, although I’m not entirely sure the discourse on there at all times is that worthwhile — “F’N GO BROWNS! JOHNNY FOOTBALL!” Regardless, it’s a new twist.
They’re getting started with tomorrow night’s opener (Packers @ Seahawks), giving the Seahawks the nod. (Not a very tough call; they don’t lose at home.) According to Quartz, it’s looking like a 74.2% chance the Seahawks will win the opening game. (Those are fairly high numbers; I would have figured lower, but still over 50 percent.)
David Rothschild is kind of one of “the guys” at Microsoft in terms of this predictive analytic content; here’s a profile of him, which is also linked in the first link above (the original story I wrote on Cortana).
By the way, if and when the Seahawks do win that opening-night game — the Packers should remember this is all their fault.