Tuesday, September 11, 2012

Bill Barnwell wrong on 2 pt conversion

Bill Barnwell kicks and misses with his best Charlie Brown imitation:

It's pretty easy to pick on Browns coach Pat Shurmur this morning, so let's spare him the cheap jokes and get to the facts. When D'Qwell Jackson picked off Michael Vick and took the return back to the house for a 27-yard touchdown in the fourth quarter of Sunday's game, the Browns took a 15-10 lead before attempting the conversion. Shurmur sent his kicking team out there and picked up an extra point to go up 16-10. On their final meaningful drive, though, the Eagles scored a touchdown and picked up the deciding score on the extra point, winning 17-16 after Brandon Weeden threw his fourth pick of the day.
This isn't an egregious decision because it came back to haunt the Browns; it's a critical failure because Shurmur chose the option that added virtually nothing to his team's chances of winning.2 Kicking the extra point gave the Browns a 16-10 lead with 14 minutes to go; the only advantage it gave them was having the ability to tie if Philadelphia kicked two field goals. That's far less likely to occur than the Eagles scoring one touchdown. The value added by a successful two-point conversion is significantly greater, more than enough to justify the risk of going for two. The footballcommentary.com two-point chart suggests that the Browns should have gone for two unless their chances of converting were below 24 percent, a conversion rate that even the league's worst rushing attack would find attainable.
Furthermore, it's an awful decision because it employs exactly zero foresight. You don't need to be thinking about win probability models or game theory to realize that going from a five-point lead to a six-point one in the fourth quarter is basically worthless. Coaches have charts that tell them when they should kick or choose to go for a two-pointer, but a second-generation coach like Shurmur should have easy decisions like this instilled in his DNA. There are some two-point decisions that require a closer consideration of the variables than the simple numbers indicate. This wasn't one of them.

Homer Smith, perhaps the greatest offensive mind in the history of college football, disagrees.  Coach Smith's formulation is simple and sound.  Going for two is the riskier strategy with a lower expected return.  It therefore makes no sense to go for it unless a coach knows that the single point has no value.  

First, compare the likelihoods of making 2 vs. making 1.  The Browns will make the kick 99% of the time -- giving an expected return of .99 points.  The average NFL try for two points succeeds less than half the time -- giving an expected return which is less than the return expected from kicking.  Additionally, the try wouldn't be attempted by the NFL average.  It would be attempted by the Browns' offense, which struggled all day and had a much less likely chance at succeeding in going for it than the NFL average offense.

Second, look at the time remaining.  Were the Browns desperate?  Did they know that the single point would serve no purpose?  Of course not.  There were all kinds of potential scoring combination outcomes for both teams with almost a full quarter remaining to play.  E.g. if the Browns go for it and likely fail, they lead by 5.  If they should later kick a FG to increase the lead to 8, the Eagles are still only one score down.  But if they kick the XPt, a subsequent  FG gives them a 9 pt lead and puts the Eagles 2 scores down. Huge difference.  And the likelihood of getting another FG with nearly a quarter remaining is not small.  The reality is that the point combinations are many.

Smith advised that the decision should only be made to make the riskier play when the time remaining was small enough that the coach can be fairly confident in predicting the likely scores he needs in the time remaining. 14 minutes is way, way too much time left (unless a coach is down by 3 scores e.g. down 24-0 and scores a TD with 14 mins left).

A final note -- using NFL averages to produce some likelihood chart on scoring by each team in a given period of time is ridiculous.  The game isn't played by an Eagle team with an NFL average offense, defense and kicking game vs. a Brown team with an NFL average offense, defense and kicking game, in average weather and average field condition.  Coaching real human beings in a real game with real differences in abilities is not remotely like crunching numbers about averages with a computer.  Games aren't played by numerical averages.

No comments:

Post a Comment