It is very often the case that a statistic, or a collection of statistics, is peripheral to the issue it examines. What quantifies the momentum of, say, Indy in the fourth quarter of this game? What might such a decision do to the morale both of my opponent’s defense in this immediate circumstance, and my own defense later? Instead of disdaining Dungy’s cautious and conservative response, based on years of experience and a respect for these kinds of intangibles, perhaps the stat man should try understanding it. Seasons are lost by decisions like these, and football teams to the coaches who make them. Where is the discussion of risk/reward? This play may cost the Patriots a Superbowl appearance, by putting them on the road in a play-off game. It’s like being offered an 80 percent chance to double one’s money–it’s much more significant if we’re trying to make a billion by losing a billion. The risk so far outweighs the reward (I have a billion already), it’s absurd to even consider it. How many chances has the Patriot defense to stop the colts from 70 yards vs 30 yards? Really, the relevant factors are only partially quantifiable; and the subtle ones not even recognizable. Myself, I neither approve nor question Belichick’s decision–what’s the point? I don’t think I would have made it, but who am I?

old man river

No correct analysis shows that Belichick decided right. Those who claim he have fallen for faulty, misuse of statistics.

The correct statistical analysis shows that Belichick was wrong. The only relevant data from which Belichick should have made his decision comes from that game alone and never from any game played in the past.

To learn why Belichick was wrong, 100% wrong, check out Gridiron Grotto http://bit.ly/1MAvQL

Burke’s probabilities refer to the longterm,but there is no longterm here. The situation is so specific it’s unlikely to be repeated much for New England in the future.

When you’re looking at single events or a limited number of similar events the average probabilities aren’t very useful.Instead you have to look at the variability of the play options available.

The punt has very low variability,most of the time the ball will be with Indy around their own 30.If you punt you almost always end up with a field position that’s very likely going to be close to the average probability for the punt.

The “go for it” option most likely gives you the ball with Indy at the New England 28 and a WP of 47% for the Pats OR a NWE 1st down and a WP of around 100%.That’s a very high variability around the average WP of 79% for the Pats if they go for it.

Billy Beane puts it best in Moneyball where he says “My “stuff” doesn’t work in the playoffs.In the playoffs it’s just luck”.

Replace “the playoffs” with “small sample size events” and he’s describing the way everyone’s tried to apply Burke type probabilities to what Belichick did on Sunday.It doesn’t work and you have to go instead with the variability.If you pick the high variability,”go for it” option you’re trusting to luck that you end up not converting……just like Dungy said.

Jump for joy because can see how all of the Belichick apologists get demolished on the Gridiron Grotto: http://bit.ly/1Q1m47