According to Adam Schefter, the Pats are 12 of 16 on 4th down attempts against the Colts since 2000.

Since Brady's first season, Pats have converted 59.1 percent of fourth downs in their own territory -- 13 of 22.

Since 2001, the teams with the two best 4th down conversion rates are Colts (61.6 percent) and Patriots (60.5 percent).

Worth pointing out that, since 2001, on 4th and 2 or less, the Patriots had converted 76.4 percent of fourth-down plays.

As if there isn't uncertainty in punting it either?

Some choice comments on the NYT blog:

Epi,
Not as much since KC prob jut runs it out if Oak gets a decent punt, which is very likely considering how good Lechler is.

Here is a gem of a comment left on the Jeff Ma site, and seems quite indicative of how well-informed people are on this matter:

Now how can you argue with that?

odelllllllll

This whole sentiment is silly to me too and its what Bruschi and these former players are saying. Well you're a professional freaking athlete, and you take it that your coach doesn't respect you, so what now, you're gonna pout and play worse for the rest of the year? How would this not motivate you to prove him wrong the next time? (Ignoring the statistics of this situation) he's just supposed to 'respect' a defense that he doesn't actually respect, at the expense of potentially losing a game?

The quote above is such a loser's mentality and made me lose a ton of respect for Bruschi and former players saying it.

People thinking it is ******edly obvious are tilting me. I've crunched various guesstimates given the circumstances and can't see either decision being too far from 0.70 WP. I think going for it is slightly better, but show me numbers that make it obviously so rather than proclaim it (and make the same assumption error you are railing against the media for).

My updated best guess:

P(convert 4th) = 0.6
Rationale: 2 yards is no gimmee on the road there, even with Brady, 3 of every 5 seems about right imo.

P(win if convert) = 0.98
Rationale: <100% because still need a 1st down to run out clock, 49/50 times seems reasonable.

P(lose if stopped) = 0.63
Rationale: I think Peyton will beat you somewhere between 2/3 to 3/5 times from inside his 30, with a timeout, and 2 minutes to run, at home with momentum -- he's within striking distance of end zone on every snap, with no timeouts I will assume probability of Pats coming back if scored on is negligible or already included.

P(win if punt) = 0.7
Rationale: With only 1 TO and having to drive for a TD, this is going to be tougher than historical averages, even with Peyton. Defense can bend as long as they don't break. Have to move chains a few times before even getting within striking distance of end zone. Not easy task. 3 TOs or NE inside 15 and it's a different story.

Thus,

P(win|go4it) = 0.6*0.98 + 0.4*0.37 = 0.736
P(win|punt) = 0.70

Going for it still looks right, but not a lock. I can see a case for all else equal but 4th down conversion here dropping to 55% (6/11 times). This would make it equal EV.

But it wasn't "or less" here, so you get a ton of 4th and inches in that sample. The distance was just long enough to pretty much take the QB sneak and run up the middle off the table. Taking those plays off the table on a 4th down play should significantly impact probabilities.

Yes, there is -- I'm not calling the Belichick decision BAD. What I'm saying is that the mathematical calculations in this thread prove nothing, because the EVs given are swamped by the nonlinear combination of the uncertainties. With only 5% uncertainty in the probabilities, I just showed that decision given by one set of parameters is statistically indistinguishable from the decision given by a different set. If we call the "null" or default model in this situation punting, we can't reject it on math alone. Appealing to Belichick's instincts and intuition as a Hall of Fame football coach is a much better justification. The point is, the call is just not obvious.

This is a fascinating issue because it is THE sports topic of the moment and helps to crystallize the difference between 2+2 world and the rest of the world. Also, it makes me want to start betting sports against the crowd, but I don't have the time for the level of analysis that would be required.

You have to use the Pats conversion rate, not the league average conversion rate, and you have to use the Colts' scoring probability, not the league average. This significantly influences the math. It's a no-brainer.

i am honestly very surprised by the backlash this call is getting. i had always dismissed the argument that the media would jump on these types of decisions or that job security was at stake. i always just thought that statistics are pretty easy and most ppl would understand.

gawd was i wrong.

I think it was the right call, but I admit it can be close if you play with the percentages. However, what is shocking is that the punting people are not arguing probability in any sense, they are not engaged in any non results-oriented analysis of the situation. They just make blanket statement like, "You don't go for it on 4th when you are ahead and on your own side of the field." That statement may be true or false, but one needs to actually analyze data to know.

Or that it boosts the morale of the offense!

My best guess:

P(convert)=.9
P(win, convert)=.98
P(win, stopped)=.20

P(win, punt)=.1

My rationale: Peyton Manning is really good at football. So is Tom Brady. I figure about 1 in 10 times Brady leads the Pats down for the winning field goal after you give Peyton the ball and he scores the go ahead TD. The rest of the time Peyton runs out the clock on his TD drive or Freeney/Mathis end the Pats comeback bid. By going for it and getting stopped you get the ball back with more time after the inevitable Manning TD than you would if you punt. So really, going for it = win/win.

Coaching is ez. I don't understand how there is 9 pages of discussion on this super obvious call.

wtf how could you possibly think this

Haven't read most of the replies but another thing is that the Colts were lucky to be able to run the entire clock down on their drive. They were first and goal from the 15 and were lucky to get Addai tackled inbounds between the EZ and the 5. This obviously helps the case for going for it on 4th because they'd have a much easier time running the clock down with a longer drive.

http://www.advancednflstats.com/2009...-vs-colts.html

advanced nfl stats has this as very obvious using historical data, not just guesses. When you factor in any team adjustments for the potency of the Pats and Colts offenses respectively, it leans even more in favor of going for it.

ya man i dunno. ive even made a bunch of posts diminishing the media effect in older 4th down discussions. i really really overestimated the comprehension of the vast majority of ppl.

BB did **** up a few things.

1. TO usage was horrible, not having any kept him from challenging plus you need them the times the Colts score and you try to come back.

2. No plan given that you're going for it on 4th. Power run on third and Brady sneak on 4th is so much better than what they did.

3. I think you need to let the Colts score once they get down that close.

and that didn't even factor in the better chance of having time remaining with a shorter drive

Yup, BB clearly had no plan with what to do if they failed on third down.. and that certainly cost them a non-negligible amount of EV.

Guy in the media who liked the call:

http://sportsillustrated.cnn.com/200...ml?eref=sircrc

I doubt if they get into the math, but my guess is Kornheiser and Wilbon would at least get the concept of going for it - probably because they get into an hour long discussion with Reali prior to the show. I really wish this was a Lebatard day, I think he's the one sports personality who might actually get and talk about how the "percentages" actually support going for it.