Patriots vs. Colts and data vs. intuition

I was shocked when I watched the ESPN highlights of the recent Patriots-Colts game. With 2 minutes left in the game -- leading by 6 points -- and with the ball on their own 28, Pats coach Bill Belichick decided to go for it on 4th-and-2.

I've always disliked Belichick and so I was happy to see the Pats fail to convert on the play, and then look silly giving the ball to Peyton Manning with less than 30 yards to go for a touchdown. The Colts scored a TD and won the game, 35-34, while Belichick looked up at the sky.

But as it turns out, Belichick's call was the right one -- if you go by the data and probabilities. As one Fifth Down blog post in the NY Times puts it, "Twenty years ago…..case closed. It was a bad move. The people with the pens and microphones had spoken. Heck……ten years ago this would have been the case. But, a funny thing happened. The N.F.L. stat geeks ... fought back."

Here's a leading geek's analysis, quoted in another Fifth Down post:

Scenario 1: The custom case for the specific offensive and defensive features of the Colts and the Patriots.

Going for it: 77.3% (Probability of Winning for the Patriots)
Punting: 75.7%

Scenario 2: The case for two N.F.L. average and equal teams in every offensive and defensive category:

Going for it: 78.6% (Probability of Winning for the Patriots)
Punting: 76.4%

Scenario 3: The break-even point on the decision occurs when the team with the ball is about 5 percent weaker than N.F.L. average on offense and 5 percent better than N.F.L. average on defense, while the opposing team is 5 percent better than N.F.L. average on defense and 5 percent worse than average on offense.

The results of Scenarios #1 and #2 clearly point in favor of Belichick’s decision, although not by nearly as wide a margin as we might have expected. Additionally, the analysis in Scenario #3 really cements the case for “going for it.” Applying this benchmark and comparing it with the far different characteristics of the Patriots and the Colts makes the call all the more clear.

Thumbs up to Belichick on a courageous and correct call last night.

In his Freakonomics column, Steven Levitt praises Belichick not only for making the right call, but for doing so knowing that "if it failed, he would be subjected to endless criticism." In other words, he cares more about winning than his reputation.

For fans of data-driven decision-making -- in business, sports, or any other sphere -- this is a fascinating case study, and I find myself liking Bill B. just a little more these days.