There’s been a lot of talk[note]More negative than positive recently.[/note] about the Browns analytics-driven front office, yet on Sunday we saw an example of how their coach Hue Jackson did the traditional thing and passed up an easy decision – from an analytical perspective – to go for to 2 in the fourth quarter.
No one blinked an eye when Jackson called for an extra point after cutting the Bengals’ 14-point lead to 8. Going for 2 in that situation probably wasn’t given a passing thought by most watching the game.
There have been posts about why you should go for 2 if you were down 15 points before scoring a touchdown, with the theory being that you want additional information (i.e. if you convert or not) earlier rather than later. This thinking hasn’t become conventional wisdom, but there has been some adoption from NFL coaches.
What’s interesting is that the case for going for 2 after being down 14 points is even more compelling, yet there was not a single instance of a team doing so in 27 fourth-quarter opportunities in 2015 and 2016.
A Simple Example
Let’s go through an example to show why there is a strong case that you should go for two when scoring a touchdown after trailing by 14.
For simplicity’s sake, I’ll we’ll make the expected extra point conversion rate at 100% and 2-point conversion rate at 50%. We’ll also assume you score two touchdowns after trailing by 14 and your opponent scores none, and that your chance of winning in overtime is an even 50%.
Your expected win probability under the traditional scenario of kicking two extra points is fairly easy to envision: you tie the game, go to overtime and then have a 50% chance of winning.
Your expected win probability going for two after the first touchdown is a little more complicated, which I illustrate below.
While the aggregate expected points in the go-for-2 scenario is equal to that of kicking extra points, you tilt the odds of winning in your favor by lumping the two missed points from a double failure into the 25% probability at the bottom, and spreading the +1 of converting the first 2-pointer and then kicking the extra point into 50% probability at the top.
The total winning probability is now 62.5% when you go for 2 (50% from making the first conversion, 12.5% from splitting conversions and winning in overtime), versus only 50% when you kick extra points.
If you use the actual NFL average conversion rates for extra points and 2-point conversions of roughly 95% and 47.5%, respectively, the boost you get from going for two doesn’t change: 12.5%, or 60% versus 47.5%.
You might think that all the assumptions going into this exercise don’t capture the reality on most teams. However, FiveThirtyEight’s Benjamin Morris ran the actual win probabilities for all potential go-for-2 scenarios and found that it’s in a team’s interest to go for 2 down 8 (scoring a touchdown after trailing by 14) at any point during the game. The benefit of going for two after trailing 14 increases as the game goes on, making Jackson’s decision to kick the extra point in the fourth quarter an obvious mistake.
Why aren’t teams doing it?
The easy answer is that coaches are acting on tradition, and that tradition usually pushes coaches to maximize the time that you still have a chance of winning over maximizing win probability.
Yes, going for the extra point after both touchdowns minimizes your chance of losing in regulation, but you still have to win the game in overtime. The failure to properly weight the probability that you lose in overtime is also a common coaching mistake.
But the Browns Aren’t an Average NFL Team
I wouldn’t argue with you that the Browns likely have a lower chance of succeeding on 2-point conversions than the average NFL team,[note]Though Kizer having mobility helps.[/note] which lowers their chance of winning in the go-for-it scenario. But the Browns also likely had a lower than 50% chance of winning in overtime, which has a much harsher negative effect on the win probability if you kick extra points.
Let’s estimate that the Browns only had a 35% chance of succeeding on the 2-point conversion versus the 47.5% league average. Let’s also estimate that the Browns only had a 35% chance of winning in overtime. With those estimates, going for 2 is still superior, though the benefit gained narrows to roughly 7%, or 42% versus 35%.
A good rule of thumb for this decision is that you should go for two unless you believe the gap between your probability of winning in overtime and succeeding on the 2-point conversion is more than 12%. I have trouble thinking of many cases where a team would be so much more likely to win in overtime than covert the 2. This would be limited to very poor offensive teams who still have a good chance of winning in overtime due to their strong defenses.
NFL is Still in the Analytics Dark Ages
Some of the talk about how analytics is changing the NFL is true, but we still have a long way to go. If the head coach of the most analytically forward-thinking team isn’t picking the low-hanging fruit of going for 2, there is clearly a lot of work to be done.
This isn’t an issue of coaches being uninformed. The Browns and other teams have analytics personnel specifically assigned to walk head coaches through these scenarios. It’s more likely that coaches fall back on their guts and tradition when the stakes are high, at least until there is enough public questioning to make them change their minds.