According to an AP story written last night, the NFL is considering a change to its overtime system, although only for playoff games. A change from the sudden-death format is overdue. Brian Burke at Advanced NFL Stats.com has found that in recent history, the team that gets the ball first in overtime wins 60% of the time. A model from Football Commentary put it at 57%.
The proposed rules aren't ideal either. In the proposal, the receiving team still wins if they score a touchdown on their first possession. If they only score a field goal, the other team gets a chance. If that other team scores a touchdown, they win. If they only score a field goal, then the game continues, and the first team to score after that wins the contest. I thought I'd run through a very quick and dirty comparison of the two systems, focusing on the change in probability that the receiving team wins the game.
As I see it, the main weakness of the proposed system is what happens when the teams score the same number of points in each of their first possessions: the old sudden death rules start to apply again. As the original receiving team has the first opportunity to score after 2 equal possessions, they possess an advantage after this point. Thanks to the model, we can estimate this advantage at 57% (this would be a heavily flawed figure if the possibility of a tie were in play, but because playoff OT just means switching sides at the end of a quarter, it's not too bad in this situation).
This already puts the kicking team at a disadvantage, one that it needs to make up by being at an advantage in the first two possessions. There is indeed a strategic advantage here: when the team gets the ball, it knows how many points it needs to either win or stay in the game. In the NCAA overtime format, teams get the ball at the opponent's 25, and the team that scores the most points after both get a chance wins the game. In Football Commentary's analysis of the NCAA rules, it found that the team that got the ball later enjoyed a win probability of around 52%.
This figure would still likely not be enough to negate the 57% advantage, even accounting for the fact that the game would usually end in the first two possessions. But it gets worse: in the NFL format, teams would likely start well short of their opponents' 25, decreasing the chance that the knowledge of the other team's points would be useful. And the receiving team's opportunity to end the game on a touchdown adds about 2.3% to their win probability, assuming the post-second-possession advantage was 57% and that second team would have had about a 20% chance of scoring a TD. It's very likely that the proposed overtime not only fails to properly account for the receiving team's post-first-two-drives advantage, but also gives that team an advantage in those drives.
When we throw in the system's likelihood to cause more ties if it were ever implemented in the regular season, it seems to me that the proposed scenario has few positives. It only negligibly improves upon the problem that it's trying to fix, and in doing so it makes the rules more complicated. Here's hoping that this proposed change gets struck down in favor of one that is simpler and actually does the job. As mentioned in the Burke article, Prof. David Romer's suggestion of moving kickoffs to the 40 yard line is a very good idea.
Here's another idea that no one seems to be entertaining: why not let the home team automatically win the coin toss? In the regular season, it would make the crowd happy, and in the playoffs, it would further reward teams for having a better year. In the Super Bowl, the winner of the toss would be the team that would have had home field. I don't like it as much as Romer's, but it's still probably better than the one being proposed.
Note: This post originally appeared on the website of the Harvard Sports Analysis Collective. Bookmark us!