<
>

# Numbers don't back up Patriots' OT strategy

With the game tied at 20, New England Patriots coach Bill Belichick chose to kick off to start overtime against the Jets. New York took the opening drive 80 yards for a touchdown to end the game, hurting the Patriots' chances of home-field advantage throughout the playoffs.

There may be situations in which choosing to kick off is the percentage play, but in normal circumstances, the receiving team in overtime will win over half the time, even under the new rules installed for the 2011 postseason and 2012 regular season. Several different ways of analyzing the question (Markov chains, backward induction and simulation) all agree: It’s generally better to start with the ball.

Our latest win probability model estimates the recipient of the opening kickoff in overtime will win about 53.8 percent of the time, and setting aside ties, the kicking team will win about 46.2 percent of the time. Deliberately choosing to kick off without any overriding consideration would cost a team a 7.6 percent chance of winning.

The raw numbers since 2012 appear to support the advantage for the receiving team, although the sample size isn’t big enough yet to be certain. Receiving teams have won 33 of the 65 overtime games that did not result in a tie, a 50.7 percent rate.

Belichick previously employed the defense-first strategy in overtime. Against the Denver Broncos in 2013, the wind was so strong and the temperatures so cold, he chose the direction of play after winning the coin toss rather than whether to kick or receive. It worked out in the Patriots’ favor then, as they beat the Broncos on a field goal with plenty of tailwind.

One argument that would support starting on defense was that both offenses were relatively ineffective during regulation. I don’t think this is a strong enough consideration to override the general advantage of starting with the ball. First, each team gets only a dozen or so meaningful possessions during a game, and so any one game’s performance is a small sample of an offense’s “true” capability. Second, the teams each scored nearly the league-average number of points. It wasn’t as if the game was tied, 9-9. Lastly, even a poor offense has a positive scoring expectation past its own 20.

In other words, even for teams with bad offenses, it’s generally better to receive the kickoff.

Given the winds and perhaps some knowledge Belichick had of his offense’s injuries and other factors, perhaps the error didn’t cost as much as a 7.6 percent chance of winning. But choosing to kick likely cost the Patriots an important advantage.