When perusing SB Nation’s NFL picks for week one, I noticed a trend when it came to the Atlanta Falcons at the Chicago Bears game. All eight of their NFL experts went with the Falcons over the Bears, which is generally what most national NFL analysts had for their pick this week.
But when I scanned all the way over to the last pick listed, I spied one brave soul that actually picked the Bears to upset the defending NFC Champs. To my surprise, it seemed that the pick wasn’t made by an NFL analyst, but rather it was made by a soulless computer. The Odds Shark computer to be more specific, and then upon further inspection, the data science team over at Odds Shark referred to their computer as SuperComputer. That’s very science-fictiony of them.
“Basically because of the stats that the computer uses to make predictions,” Odds Shark explains, “it doesn’t trust the Falcons after their Super Bowl LI performance.”
I see their artificial intelligence subscribes to the old Super Bowl hangover theory.
They have the Bears winning on Sunday with a predicted score of 24.4 to 22.5.
So scientifically speaking, the Bears have a chance?
I needed to know more about SuperComputer, so here’s what I discovered.
The SuperComputer uses a selection of mathematical models trained on over 30 years of offensive and defensive statistics. It looks at historical betting lines and how they relate to the overall outcome of a game. Our models are constantly retrained to update assumptions about what underlying stats really matter for that final score.
If science and math are on the Bears’ side, maybe the WCG writers that picked Chicago to win are a little more analytical than those of us that didn’t.
I had to get even more info on this, so here’s the reason they gave me for why the SuperComputer hates the Falcons in this game.
Even though it accounts for years of previous games, the SuperComputer weighs recent performances more highly because the sport is always changing. Matt Ryan and company may be ready to move on from their SB51 performance, but the numbers haven’t forgotten. Atlanta allowed 37 first downs while achieving only 17 and their time of possession was skewed by a factor of 2:1 in New England’s favor. Compare that to the Bears’ last game against Minnesota — the score may have been a blowout, but downs and possession were almost dead even. This is a simplification of the computer’s “thought process”, but it illustrates how the behind-the-scenes numbers are considered when looking at previous games.
I may never understand their complete statistical analysis, but
anyone anything that picks the Bears is alright by me.