Going For Two

By Robby Wellington – Staff Writer

Today I am coming off a long hiatus to talk about GOING FOR TWO!

Yesterday, Dalton brought up the issue of 2-point conversions and loyal RotoScoop reader, Chad, had the following to say:

“One last thing though Dalton…I’m not sure if you estimated those success rates on the Extra Pt vs. 2 point conversion, or if you pulled them from a stat somewhere…But obviously every team has a different success rate. But the expected value of those two is indeed pretty close.

Expected Value of Extra Pt. = 99*1 = .99
Expected Value of 2pt Conversion = .45*2 = .90
Difference in expected gain is only 9 hundredths of a point.

I actually use this example when tutoring my students in game theory. But I usually say, if the extra point is 96% likely (which I believe is the correct stat), what success rate at 2 point conversions would you need, to make attempting a 2pt Conversion every time to be feasible? The correct answer would be 48%. In the 1990s, the average across the league was indeed much less than 48%, but over the past two seasons it is more than 50% and has been climbing for the last few years. Obviously I understand the value of “nearly guaranteed points,” but as a statistician, I prefer the idea of maximizing total points scored over the game. At certain points in the game, there would be no reason to go for two, because the added bonus of the two point conversion may provide you no additional gain. The most obvious example would be a tie game in the fourth quarter. So in those cases, the statistics don’t apply, but in the first half, going for two seems like the better choice.

Also when down two touchdowns late in a game (10 mins or less left in 4th) YOU SHOULD ALWAYS go for two on the FIRST score. No matter what, you’ll need two touchdowns. If the attempt is no good, you can still tie on a second score with another 2pt attempt. If the 1st attempt is good, an extra point on the second score wins the game. Conversely, if you choose to kick the first time, you are committing to overtime at best. Unless some crazy coach wanted to go for two on the final score, to be faced with a win/lose scenario riding on a 2pt conversion.”

For the most part, I have to agree with Chad. First off, I have been talking about going for two when cutting a late deficit to eight points (i.e. scoring the first of two touchdowns to tie a game) for years now and am dumbfounded when no one ever does it (Check out this article for more; it looks like some bozo in Montana ripped off the idea).

This move is almost as critically underutilized as letting a team score a late touchdown to go down eight instead of letting them run out the clock, which is what Joe Gibbs presumably did on Sunday.

Back to two-point conversions, I do actually think teams should go for two a bit more than they currently do. Since the inception of the two point conversion in the NFL in 1994, conversion percentages have hovered in the mid to low 40s, with the number climbing up above 50% for the past two years as attempts have declined. Without getting too caught up in statistics (determining what counts as “statistically significant,” etc.), I’ll just go ahead and say that the NFL average conversion rate in a close game is probably slightly less than 48%, which is half the 96% success rate of PATs.

This means an average NFL team in an average situation against an average defense should kick the PAT.

However, there are two scenarios where I think teams should go for two. One is where they feel as if they have a 50% chance or better of converting, whether the offense is on a roll in the game, the defense is terrible, the Pats are playing (seriously, why hasn’t Belichick gone for two every time this year?!), Plaxico Burress is matched up on a 5’9” cornerback; basically any scenario when a team feels they have a better than average chance of converting. This is, of course, excluding obvious situations where a team is down one point, or seven points or three points later in a game.

Which brings us to the other factor in determining whether a team should go for two or not: the score and time of the game. Take Sunday’s Eagles vs. Redskins game. Let’s go ahead and assume that both team’s had the same chance of converting the two, and that this chance was slightly less than half their odds of converting a PAT. I still credit the Eagles with going for two but fault the Redskins for the same tactic. Why? Well, the Redskins went for the conversion in the first half, in an effort to go up seven points instead of six. That early in the game, it is virtually impossible to determine what is going to happen to the score later and an early lead of 6 points is certainly not the same as a 5 point advantage. In fact, the advantage of 7 points over 6 is no greater than that of 5 points over 6, right?

On the other hand, the Eagles went for two to tie the game late in the third quarter. Sure, they went down nine points in the fourth quarter, but they would have still been down eight and would have still had to get a conversion later in the game. The disadvantage on not converting and being down two was not nearly as great as the advantage of converting and tying the game relatively late in the contest.

Anyway, it is late now and people are carrying a full-on conversation in front of me, clearly oblivious of my blogging effort. But I promise to write something good soon, and it won’t be entirely theoretical. Thanks!