At Edj Analytics we often discuss the risk aversion biases of NFL coaches and how it can negatively impact the outcome of a game.   The most costly decisions often arise during short yardage 4th down decisions in the red zone but this is not always the case.  Last weekend when the Washington Redskins faced the Philadelphia Eagles there was a particularly interesting situation late in the 4th quarter.

With 1:53 remaining in the game, Washington handed the ball to Chris Thompson on the Philadelphia 25 yard line and he negotiated a path all the way to the end zone.  A jubilant Redskins team now had gained control of the game and left the Eagles with the tall order of marching the length of the field and scoring a touchdown with only one timeout and about 1:50 of precious clock.  What could possibly be wrong with this picture?

Surprisingly, the Redskins made a clear error.  There isn’t a running back in the NFL who would want to hit the brakes in the middle of a touchdown run, but that is exactly what Chris Thompson should have done. If he could have taken a knee inside the Philadelphia 10 yard line he would have increased his team’s winning chances significantly.  Let’s break down this situation.

By taking a knee short of the goal line, Washington can run the clock down to less than 25 seconds and give themselves the equivalent of a (traditional) extra point to take a two point lead.  This strategy would provide them a resulting win probability (GWC) of approximately 88% – as they should convert the field goal about 98% of the time and the Eagles will fail to score (a field goal or better) on the subsequent desperation drive approximately 90% of the time.

By Chris Thompson running in to the end zone for a five point lead, Philadelphia will now be required to score a touchdown on their last drive and will have about 1:50 to do so with one timeout.  Washington will of course attempt a two-point conversion (which they did and failed) and will have a lead of seven about half of the time.  According to our simulation model, in this situation the Eagles should score a touchdown on their last possession about 20% of the time.     Half of the time that touchdown will result in a win (when they are down by 5) and the other half of the time (when they are trailing by 7) the touchdown will get them to 50% GWC in OT (or 50% with a two point conversion) resulting in a win probability of approximately 15% for Philadelphia and subsequently 85% for Washington. Therefore Washington, by employing a very unconventional but also very executable strategy, could have increased their winning chances from 85% to 88%.  To put it in proper perspective, a single error of this magnitude in each game of the regular season would on average cost a team about half a game per season.