things kevin hates

You’ve seen it happen tons of times. A team gets the ball with a few seconds left in the 1st half and rather than take a shot with a Hail Mary pass, the quarterback takes a knee and the team trots off to the locker room. But with a few seconds left, why not take a shot? This is yet another example of bad football coaching: teams play it safe instead of making the play that would maximize their chance of scoring points (and thus their chance of winning). One example occurred in last night’s Saints-Patriots game.

The Saints opted to take a knee after Stephen Gostkowski missed a 50-yard field goal, giving the Saints the ball on their own 40 with 13 seconds left. Now, a score for either team is certainly unlikely. But common sense dictates that there is a break-even yard-line at which the defense would be more likely to score than the offense.* I certainly wouldn’t want a team to go for it from their own five yard line; the risks of a safety or turnover would be too great. The Saints had two timeouts left; a thirty yard pick up would have gotten them into field goal range. Was any score a long shot? Yes. But you’ve got the best quarterback in the NFL, he’d been tearing through the Pats’ defense, and most importantly of all, while the Saints may only score a small percentage of the time, the Patriots would score even less often.

The Saints could have, for example, left Pierre Thomas and David Thomas into block, Robert Meachem and Devery Henderson on streak routes down the sidelines, and Marques Colston twenty or thirty yards over the middle. Tell Drew Brees to avoid a fumble at all costs—fall down and take a sack if you have to—and he can either chuck a long ball to Meachem or Henderson, either getting a long catch or a pass interference call, or find Colston and hope he gets into field goal range. An interception on a Hail Mary is extremely unlikely to result in a runback for a touchdown, especially with a speedy back like Thomas in the backfield to run down any long returns. And with as little time left as there was, the Patriots would have had to score on the runback—it’s extremely unlikely that they would have found themselves in field goal range with time left on the clock in almost any plausible scenario.

If I’m going to say that the numbers dictated going for it, I should probably come up with some reasonable numbers. First, what were the odds the Patriots would score? To score they’d need a turnover. Including the Patriots game, Drew Brees has thrown nine interceptions and fumbled the ball seven times (losing 4). That’s 16 picks+fumbles in 343 pass attempts, roughly 4.7%. The Patriots have been pretty good at forcing takeaways this year, so we’ll bump that number up to 5%. But what are the odds of scoring on a turnover? The Patriots have scored two defensive TDs off 23 takeaways this year—less than 10%. Now that’s a mighty small sample size, so what are the numbers for the whole league? Using the 2008 turnover data from this page and the fumble and interception TD totals from this page, we get 793 turnovers and 85 TDs, about 10.7%, not far off from the Pats’ numbers for this year. To simplify things, we’ll say that the Patriots had a 10% chance of scoring on a turnover. If anything, 10% is high for this situation, since the Saints would have been especially careful. But sticking with it, the Patriots had about a 0.5% chance of scoring.

Now, what were the Saints’ chances of scoring? It’s harder to figure out, but I think it’s safe to say that it was well above 0.5%. Drew Brees completes more than two-thirds of his passes. Obviously it’s harder to complete long passes than shorter ones, but looking at ESPN’s ridiculously comprehensive split stats, Brees is 5-for-15 on passes thrown 31 or more yards. Even if we lower that percentage since the Patriots would have been in a prevent defense, I think 20-25% is reasonable. If we use 20%, let’s say 80% of those would be passes that get into field goal range and 20% (4% total) would be touchdowns. That doesn’t even include pass interference penalties. You could say that refs never throw a flag on Hail Mary plays, but they did just that in Week 11′s Browns-Lions game, where Matthew Stafford’s pass resulted in a 31-yard pass interference call with no time remaining, followed by a touchdown on the next play. Even if my 4% number is high, the Saints still had an excellent chance of completing a pass which would have gotten them into field goal range, or at least into better Hail Mary range. And again, remember: 13 seconds left, and two timeouts. Definitely time to get two plays off, possibly even three. And the half can’t end on a defensive penalty, so if they can get one of those, they could get an extra play. I’d certainly welcome arguments with these percentages, but the odds of a Pats score were so small that I don’t see how you could possibly say that the Saints were less likely to score.

Obviously Sean Payton wanted to play it safe with a two touchdown lead. But the Patriots would be getting the ball first in the 2nd half, meaning a touchdown drive could make it a one possession game. So more points would have been very helpful.

* Strictly speaking we need to consider the expected value of points, not the percentage of times the teams will score; with 13 seconds left either team could conceivably get into field goal range, whereas with a second or two left only a penalty could prolong the half beyond one play. But for the purposes of this point we can call it a wash.