## Belichick's 4th Down Decision vs the Colts

New England coach Bill Belichick is taking a lot of heat for his decision to attempt a 4th down conversion late in the game against the Colts. Indianapolis came back to win in dramatic fashion. Was the decision a good one?

With 2:00 left and the Colts with only one timeout, a successful conversion wins the game for all practical purposes. A 4th and 2 conversion would be successful 60% of the time. Historically, in a situation with 2:00 left and needing a TD to either win or tie, teams get the TD 53% of the time from that field position. The total WP for the 4th down conversion attempt would be:

(0.60 * 1) + (0.40 * (1-0.53)) = 0.79 WP

A punt from the 28 typically nets 38 yards, starting the Colts at their own 34. Teams historically get the TD 30% of the time in that situation. So the punt gives the Pats about a 0.70 WP.

Statistically, the better decision would be to go for it, and by a good amount. However, these numbers are baselines for the league as a whole. You'd have to expect the Colts had a better than a 30% chance of scoring from their 34, and an accordingly higher chance to score from the Pats' 28. But any adjustment in their likelihood of scoring from either field position increases the advantage of going for it. You can play with the numbers any way you like, but it's pretty hard to come up with a realistic combination of numbers that make punting the better option. At best, you could make it a wash.

### 300 Responses to “Belichick's 4th Down Decision vs the Colts”

1. Anonymous says:

This post is total garbage and meaningless without links to the data purportedly backing it up:

"A punt from the 28 typically nets 38 yards..." Citation? Link?

30% chance of scoring TD from own 34 in less than 2 minutes? Citation? Link?

That I really do not buy. The Colts have scored 30 offensive TDs, but thrown 8 INTs, punted 36 times, attempted 14 FGs, and appear to have lost 6 fumbles (cites: sports.yahoo.com/nfl/teams/ind/stats and nfl.com/teams/indianapoliscolts/statistics?team=IND). Thus, total on all possessions they score a TD only 31% of the time. Sure, many drives start from the 20 after kickoffs, but my quick calculation also does not account at all for the time of the drive -- many, many of those 30 offensive TDs might have taken far more than 2 minutes. I call BS on this.

2. Zach says:

To anonymous above, all the data he uses is from play-by-play data since 2000 or 2002--the same data he uses for all his articles.

3. Anonymous says:

Ok, I see it's really 32% of the time (30/94 = 31.91%).

Still, I sincerely doubt a number of the key stats casually tossed around in this post.

4. Anonymous says:

Then he should cite to it. This is the internet -- use links. He did for the purported 60% success rates on 4th and 2.

5. Anonymous says:

It's such an interesting argument. Years ago, no coach would have gone for it. Now several, including BB, often go for it and as we see here very often make it.

One interesting non-statisticl point is the momentum boost that Indy got from stopping their offense on 4th down deep in their own territory. I think BB - and all coaches - should have taken this into consideration.

The opposite of this would have been a booming punt and big tackle by NE special teams. (Also unlikely, I know)

6. Happy says:

I thought it was the right choice at the time. I've read several analysis defending Belichick, rightfully so. I actually thought Belichick was giving his team three ways to win instead of one (I haven't seen the third included in the analysis).

1.) NE converts the 4th down and wins
2.) NE misses and keeps Ind out of the end zone and wins
3.) NE misses and IND scores too quickly and NE drives for a field goal with the time that was remaining because IND only had to travel 28 yards .

When playing poker, sometimes you get unlucky, but you still feel OK if you're playing correctly and you know you'll win in the long run. Belichick played this correctly. That's part of what separates him. He had the confidence to do the right thing rather than caving to convention.

7. Anonymous says:

Or:
1) NE punts, and IND turns over on towns
2) NE punts, and IND throws INT
3) NE punts, and IND fumbles
4) NE punts, and time expires before TD

As I noted, none of these % have any citations or data to back them up, some are highly questionable, and this whole post is BS. It was the wrong call.

8. Anonymous says:

Why sould he have to cite himself, on his own website?

9. Jim Glass says:

Imagine if Zorn had done this, they'd have burnt him at the stake.

It really shows that sometimes a coach has to have the job security of Belichick to do the *right* thing.

Now think of the larger lessos -- sports is just a small, entertainment microcosm of larger life.

All the same processes are going on in politics, democracy, world affairs, war & peace decisions, the major media ...

Or maybe it's better not to think about it.

10. Anonymous says:

Driving to work, listening to Colin Cowherd, of all people, talking about how his callers and emailers didn't understand statistics had me cursing under my breath the entire way.

11. Anonymous says:

I'm not over-thinking anything, you're simply not thinking. Read my post more carefully, I said "given the assumption that a first down would win the game," (which many are asserting) there would be "a 60% probability of winning based on successful conversion" - since as you state there is a 60% probability of converting. If 'conversion' = 'win', then 'probability of converting' = 'probability of winning' - Get it? Clearly, all of these stats are an over-simplification, I simply want to point out that even in this over-simplified case, the author is misinterpreting the result and convoluting an otherwise simple analysis.

12. Andy Cahn says:

This is great - giving fuel to Pats fans to argue that BB made the right call, sounding like conservatives who think they won the election in NY23. Newsflash: you both lost. Therefore, the wrong call.

13. Anonymous says:

Percentages are only good when put into context for the situation. What is the percentage of teams converting 4th down within their own 30 while winning the game? Better yet, what is New Englands conversion for this situation? If you got those numbers, then we can decide if this was a great call or a blunder.

14. Kevin Camp says:

What I think is hilarious is that the people who, myself included, think that Belichick made the right call have
A) the math
AND
B)a head coach who has won three Super Bowls and is generally considered one of the greatest football minds of all time
on our side and people still tell us we're wrong.

It's like Bill Gates and Warren Buffet started a company together and people are saying "Don't buy stock in THAT company."

15. Kevin Camp says:

Andy Cahn said...
"This is great - giving fuel to Pats fans to argue that BB made the right call, sounding like conservatives who think they won the election in NY23. Newsflash: you both lost. Therefore, the wrong call"

Aaannnddd, we have a winner for the dumbest comment of the day. By this logic, BB made the wrong call in starting Brady. I mean, that's what he did and he lost, so it must be the wrong call.

16. Anonymous says:

Adding the probability of winning based on a failed conversion to the probability of winning based on a successful conversion, is convoluting as the two scenarios can not simultaneously exist. Once the decision is made to go for it, there exists a 60% probability of winning via successful conversion (assuming that converting leads directly to a win - best case scenario). If the play results in a first down, the probability of winning based on failure to convert is instantaneously irrelevant. If the attempt fails, the probability of winning instantaneously drops to 47% based on preventing a TD from that field position. At this point, the probability that they would convert is irrelevant. On the other hand, had they punted, the probability of winning would have been 70%. Ipso facto, punting would have given the best probability of winning. Thus, the appropriate way to assess the situation was not the binary, should we go for it or not (wherein we blur two possibilities into one - successful first vs unsuccessful attempt), rather the three-fold, if we go for it we have a 60% chance to win outright, if we fail to convert we then have a 47% chance to win by preventing the TD, if we punt we have a 70% chance to win by preventing the TD.

17. John says:

What would everyone be saying if they had punted the ball and they Colt marched 80+ yards for the same game winning TD?...They'd be bitching and cursing Bill saying "Why didn't he go for the win on the 4th and 2?" Foolish people....relax!

18. Anonymous says:

To the post two above me, just to clarify...are you saying that

Prob(Win | Attempt 4th&2) = 60% AND
Prob(Win | Attempt 4th&2, Fail Conversion) = 47%

Because you mention that there is a 60% chance patriots win given sucessful conversion. What are the chances that patriots win given an unsuccessful conversion? In the sample space of a patriots win, it must include the possibility of pats winning given an unsucessful conversion.
Unless that probability is zero, then the probability of a patriots win must include some component of them winning even if they failed the conversion.

19. Anonymous says:

What's even better than that is Peter King spelling out the math, and then STILL arguing that 'The Hoodie' was wrong! What a moron.

20. Hogwallop says:

Here's one thing that I haven't seen, and I apologize if I missed it in an earlier comment: The percentage of what teams have done on 4th-and-2 in the past is irrelevant.

The fact that teams have converted 60 percent of previous 4th-and-2 situations doesn't mean the Patriots had a 60 percent chance of converting this particular 4th-and-2. They had a 50 percent chance: They either converted or they didn't. After all, if I flip a coin 100 times and it comes up heads 60 times, that doesn't mean the next flip has a 60 percent chance of being heads.

So here's the relevant math: The Patriots had a 50 percent chance of converting the fourth down and, basically, guaranteeing a win, and a 50 percent chance of giving Peyton Manning the ball 30 yards away from the endzone. That's the jumping off point for the discussion.

21. Zach says:

Hogwallop, that's like saying the fact that only 5% of punts (as an example) are returned for touchdowns is irrelevant, because you either return it for a score or you don't...which is wrong.

22. Zach says:

By the way, even if the fourth down conversion was 50%, going for it is still the better decision according to Brian's numbers:

1 * .5 + (1 - .5) * (1 - .53) = .735, which is greater than the 70% chance by punting.

23. Hogwallop says:

Zach:

On any given punt, the chance of a touchdown return vs. a non-touchdown return is 50 percent. What has happened before has absolutely zero bearing on that one punt.

That's the point that I'm making. It is completely and utterly irrelevant how successful NFL teams have been on 4th-and-2, how successful the Patriots have been on 4th-and-2, and how successful the Colts defense has been on 4th-and-2. There were two possible outcomes to that one play: Conversion or no conversion. That's it.

24. Mike says:

What about the play? Indy had broken up almost the same play the previous down, and the ball was thrown just to the 30, no further. If he gets tackled backward, which is likely given that Indy will be defending the 30 like their lives depend on it (b/c they did), then you are depending on the refs to essentially consciously grant or deny you the first (they know you need EXACTLY the 30. Why not throw it further down field or run it?

25. Anonymous says:

I have been a regular reader to this site for a while now. I thought it was a correct decision for NE to go for it on 4th and 2. You have one play to win the game. Why not take the chance and not give Manning a chance to beat you?

I am glad Brian ran the numbers and that it supports the idea to go for it.

One reason that Belichick went for it was because he was in a similar situation three years ago.

In the 2006 AFC Championship game the Patriots faced almost the exact same situation as on Sunday night. In the 4th Quarter with 2:27 remaining New England was leading Indianapolis 34 to 31. It was 4th and 4 at the NE 46 yard line. Belichick decided to punt the ball and rely on his defense. The punt went into the endzone so Indy got the ball at its 20 yard line with 2:17 on the clock and only one timeout left. By the time the 2 minute warning came Indy was at the New England 11 yard line. Indy drove the ball 69 yards in 17 seconds. A roughing the passer penalty helped Indy along the way.

It took three more plays and Indy was in the endzone for a touchdown. That proved to be the winning score.

I have to believe Belichick thought about that game and that gave him a reason to go for it on Sunday night.

26. mikeinbrooklyn says:

I am shocked that teams score 30% of the time from their 34 yard line. I had thought this was a terrible move by Belichik. Thanks for showing me the truth.

Also, I just found this website now. I am really into advanced stats in baseball and had no idea there was a site like this for football. I am very, very excited!

27. Anonymous says:

Hogwallop: next Sunday the 9-0 Saints are playing the 1-8 Bucs. From the Bucs perspective, either they will win the game, or they won't. Thus, by your logic, the Bucs have a 50% chance of winning this game. If you truly believe your 50-50 logic, please contact me, I have lots of bets I would like to make. Or at least call my bookie to convice him that he should pay me even money if I bet the Saints straight-up.

28. Anonymous says:

As we have seen today in most comments around the web, the majority of people just believe in what "everybody knows." "Everybody knows" you're supposed to punt here. Also, "everybody knows" we are running out of oil. "Everybody knows" we needed to bail out the banks. "Everybody knows" the US has the worst health care in the world and we need to spend billions to fix it. "Everybody knew" housing prices were going up forever (2005) or were going to drop (2009). "Everybody knows" this or that guy is guilty. "Everybody knows" global warming is real. You get the idea. If you think this kind of don't-let-the-facts-get-in-the-way-of-what-I-think is isolated to sports, you're crazy.

29. Anonymous says:

Brian says, “You can play with the numbers any way you like, but it's pretty hard to come up with a realistic combination of numbers that make punting the better option. At best, you could make it a wash.” That may be true for the NE/Indy game but the decision to punt or not was based on facts of the game and league stats will never help answer that question. It’s based on league average numbers so what holds true for NE/Indy must hold true for say Detroit at Washington. Say the score is 6-0 Detroit, and Detroit is in NE’s exact situation, going for it on 4th down seals the win. Brian’s model says “Go for it!”, now you tell me how long that Detroit coach would last if he did that. Punting is CLEARLY the best option for Detroit.

Same thing last night, if the score is 6-0 NE punts every time. You can play with the numbers all you want and get meaningless results, but it is pretty easy to make a case where punting IS clearly the better option. The logic behind Brian’s analysis is flawed; it’s biased based on who was playing and what the score was at that point in the game. Unless you can convince me that Detroit should go for it.

KenYonLV

30. matthew tualitan says:

these numbers are great, but i question their use. what are the game situations surrounding most 4th and 2 attempts? i'd fathom --since they're already broken down to outside the 20-- probably 50% are from teams just outside of FG range where a punt is useless, and most of the remainder are in 'offensive do-or-die' situations in the 4th quarter where going for it is a necessity. i'd imagine very few of the 60% conversion rate numbers are taken from a situation where a team is deep in their own territory against a 'defensive do-or-die' scenario. in other words, the numbers are outstanding, but they assume all 4th and 2 situations outside the 20 are equally played by both sides of the ball, which i think any former player or coach would disagree with. in the end, the 4th down call was ballsy, but possibly a break even. on the other hand the clock management and play selection on the final drive was probably much more costly

31. Anonymous says:

I've been trying to explain game situations like this to sports fans for forever, it's impossible. Unless someone thinks like a logician their emotional "sports convention" defenses won't let them entertain opposing views to the "all powerful convention". People have a visceral reaction to sports decisions hammered into them over the years. (Those ideas of course having been come up with by people who are now dead because their life expectancy was much less than it is now thanks to medical and nutritional EVOLUTION).

Just like almost all football "conventional wisdom", hiding like an ostrich and punting on 4th and short, was invented 50 years ago and never reassessed.

Two of the main reasons you punt are made moot by this game situation. 1. You punt to improve YOUR expected field position when getting the ball back. Also increasing the likelihood of you outscoring your opponent over multiple possessions. (Yes, you also punt to worsen the opponents field position but that is a different reason). This situation eliminates all long term decision inputs and is easily represented above in the article. Future field position of the team with the lead is irrelevant as any first down is game over. 2. The opposing team has only 3 downs to get a first down. An elite QB with 4 downs usually moves the ball fairly easily. (Teams behind late score at much higher rate than during the game).

Coaches are still hiding behind "the conventional" despite a mountain of data all around them. Think about it from the coach's perspective, if they punt they are absolutely without blame regardless of the outcome because NOONE goes for it inside their own end. Imagine if a young bright coach without 3 rings tried this. He'd probably get fired even though the numbers were clearly in his favor. Football coaches, and fans, are slaves to convention.

I can't stand Belichick but he is the only coach, college or pro, who I've ever had to figure out what he's doing, why, and I've never been able to improve his decision making, including now. (Of course he shouldn't have spent the timeouts but I consider that a mistake, a rare one, not a decision).

The man does not take vacations. He loves working 100 hour weeks, takes pride in it and he's incredibly bright. He thought about this decision 5 years ago, not when it happened, he was merely trying to get the play right. While other kids had wet dreams Belichick dreamed about zone blitzes.

He's not embarrassed by the call he made but doesn't want to explain it to anybody and even the playing field for the convention slaves. He coaches rings around these other nimrods. I hate Belichick but I wish he coached my team.

The data quoted in this article is based upon generic teams and we all know the Patriots are better at 4th and 2 than generic teams and that Peyton Manning has better stats in game winning situations than generic teams, especially how hot he was in the fourth. (I heard on the radio that the Pats were 76 percent over the last two years on 4 and 2 or less).

All of those 4th quarter and behind stats also have a larger sample of bad or mediocre teams than good teams, because bad teams are more likely to be behind. How many Brady stats were in there the year they went 16-0. None (Giants, maybe?), but I'm sure Tyler Thigpen, Jon Kitna and Jason Campbell are all in there, quite a bit.

That skews the numbers once again, significantly, to the omnipotent hoodie.

And the statheads shall inherit the earth.

Thanks for listening.

32. dino says:

For the analsis to work your historical 4th and 2 conversion rate must be derived entirely from short passing plays hoping to be caught at the 1st down marker.You can not include running plays or even deeper passing plays.

Ideally you'd also need to restrict your data to plays that would produce a similarly extreme swing in win probability if it failed to try to reproduce the stress levels experienced by the players.

I think your data set is probably too remote from the reality of the actual play call to allow any concrete conclusion as to whether Belicick was right or wrong.

33. Anonymous says:

For the game, the colts were averaging 6.5 yards per play. They also averaged about 50 seconds to go 40 yards on scoring drives, so even at normal speeds the colts could be back at the 30 yard line following a punt with over a minute to play.

So which is more likely:
1. the patriots turn the ball over on downs
2. the colts drive ~40 yards following a punt

The historical odds of #1 for the Patriots in this situation is about 37%. In this game, they had nearly 500 yards so the odds of failing the conversion was likely around 30%.

The Colts had scored from 79 yards or greater 29% of the time in the game, but 2 of those scores were in the 4th quarter where they were averaging 14 yards per play and were only forced to use a 2nd down 3 times. One could argue the Colts were likely to win the game even if they had to go 70 yards. So the likelihood of just reaching the 30 yard line was considerably greater than 30%.

Going for it was easily the correct decision in this situation.

34. Anonymous says:

## For the analsis to work your historical 4th and 2 conversion rate must be derived
## entirely from short passing plays hoping to be caught at the 1st down marker.
## You can not include running plays or even deeper passing plays.

Brady had completed 66% of his passes for the season and 69% for the game. He also completes 74% of passes that are thrown 1-10 yards. I think a 63% conversion rate when passing on 4th & 2 is very realistic.

35. dino says:

"I think a 63% conversion rate when passing on 4th & 2 is very realistic"

Brady's below 40% this year when passing on 3rd or 4th and 2.

36. Anonymous says:

Let's say you had your life savings riding on the Colts. Seriously, try to imagine that. A genie appears and says he'll make Bill do whatever YOU want: punt or go-for-it. What would you choose?

I can't speak for anyone else, but I had some serious \$ riding on the game, and I HATED to see them go for it.

37. Anonymous says:

How about this stat and the probability that comes from it. The Patriots were previously 82-0 when leading by at least 13 going in to the fourth quarter. So the probability that they would beat the Colts:

X = number of games where the Pats lead by at least 13 going in to the fourth quarter

Y = number of wins

Probability that the Pats would win

Y/X*100 = 100%

Hopefully, this stat is more convincing. Belichek was just playing the percentages that the Pats would win regardless of his decision.

Clearly, the Pats losing is just an illusion that has you all fooled. It is impossible for them to lose as they clearly win 100% of the time in that situation. Thus those of us with the ability to cipher understand that they actually won that game.

38. gabefung says:

>Brady's below 40% this year when passing on 3rd or 4th and 2.

Well afterall he came out of a major injury break, it is understandable his performance in early season suffer. Maybe a recent 4 games stat would be more suitable to compare?

39. Anonymous says:

Perhaps going for it on 4th-and-2 was the right move and the wrong move was a shotgun formation with a pass to Kevin Faulk (which the Colts said that was one of two people they thought the play was going to) that was delivered right around the first down line.

40. Jaime says:

The problem with ESPN and other critics of the play is that they are judging the call based on the result.

It is easy to judge after the fact. Had the Patriots converted, they would've been saying what a great coach and strategist he is.

41. Anonymous says:

Re: Trent Dilfer, Rodney Harrison and Teddy Bruschi

the idea that going for it on 4th down damages the defense and thus team unity- the coach is telling them he has no faith they can stop the other team's offense: what message does the coach send to the offense every time he punts on 4th down instead of going for it?

42. Anonymous says:

Bill is a math genius to do all those calculation in his head. If you aska politician what he/she would do, the answer may be "If the team go for 4th and 2 and loses, the coach get the blame. If the team punts and loses, the defense sucks." So Bill is a math genius and a bad politician. He still gets the blame anyway.

43. Anonymous says:

What about taking a safety to slightly improve the punting situation. A safety leaves the Pats up by 4.

44. Anonymous says:

I have a few observations...many have pointed out something along the lines of "Oh well Bill showed faith in his offense by going for on 4th down." What message does that send to the defense though? That they can't manage to hold the Colt offense one last time from driving down the field some 70 yards?

And does this take into account that the Patriots were the away team?

What about the median and mode of average yards gain in this situation? In other words, sure a running back can average 4 yards per rush, but take out the 45 yard breakaway and it drops considerably.

I feel like most people posting here are the sort of obnoxious Boston fans who will defend their team to the death. There is NO WAY that Bill knew the odds when he chose to go for it. Where would the cut off have been? On their own 20? On their own 10? At the 5? I mean even at midfield it seems an interesting option of peggin the Colt offense inside the 10.

Next question: What if it was 4th and 3? Again where do you draw the line here? Go for it on 4th and 2.5?

Again, Bill did not know the stats, which admittedly according to your own site seem to defend Bill a bit. But who here REALLY thinks it was a good decision? Or how about the call you throw a pass on 3rd and 2? Or even the play call to throw on 4th and 2 instead of running the ball twice? Bill made a number of mistakes late in the game here.

45. Anonymous says:

This is really a case of inaccurate stat work. First of all taking NFL baselines, which give some true indications, are somewhat meaningless for a variety of reasons:
1. You have to consider the Defense you are going against,
2. You have to consider WHETHER OR NOT YOU HAVE THE LEAD!!!
3. You have to consider your own personnell
4. You have to consider the point in the game at which you are trying to go for it on 4th and 2. Is there a good sampling of stats on 4th and 2 with a 6 point lead with 2 minutes to play???? Certainly this is a different situation than with 10 seconds to play or 10 minutes.
5.

It's analagous to this: Would you bet your mortgage or life savings on a horse who's got a 66% chance of winning??? yeah the odds are in your favor but not overwhelmingly,is it worth the risk?
Going for it on 4th down when you are behind is worth a gamble because the alternative could be certain defeat...that was not the case here. You had the win. Factor something else into this as well. What would the odds be if Belichik ran the ball 3 times and then ran the playclock down to 1 second before he punted. Manning would have the ball on his 35 or thereabouts with 1:10 left on the clock and no timeouts needing a touchdown. What are the odds of that happening?

46. Anonymous says:

I wonder whether we can agree on one particular part of the equation. Forget league averages - after watching the entire game, what do you think the chances were of Manning scoring from, lets say, his own 35 after a punt with 1:55 on the clock and one timeout? My estimate is 40%

47. Anonymous says:

Anyone else think the spot was pretty horrible? Looked like forward progress and a first down to me, or at the very least a lot closer than the spot.

48. Anonymous says:

I knew the moment Belichick went for it these types of calculations would be presented. I would suspect your conclusion to be correct but your formula is woefully inadequate to properly measure the variables associated with this problem. As has already been previously pointed out one cannot equate the likelihood of the Colt's scoring a TD from 30 yards with the likelihood of the Colts scoring a TD from 70 yards. Further to consider all 4th and 2 conversions to be created equal is to be misinformed and make an unrealistic assumption. To consider the Patriots D and the Colts offense to be average (or actually the mean) and vice versa is a broad supposition. I doubt their is a data set large enough for each team and circumstance to come to a reasonably well informed conclusion on actual percentages, I would use regression analysis but your conclusion is only as strong as your data set, which in this case I would guess as weak. That said I would guess the percentages are in Belichick favor

49. gabefung says:

Alternatively, consider:

1. You have to consider the Offense you are going against (after punting the ball),  2. You have to consider how efficient their offense are (after punting the ball),
3. You have to consider your own personnel (after punting the ball),
4. The big reward after taking this big risk

>after watching the entire game, what do you think the chances were of Manning scoring from his own 35

the Manning in first, second and third quarter would be 30~35% for me. Patriots defense played well and keep him 3 and out.

but the Manning in forth quarter would be *at least* 50% for me - in 4rd quarter, he strike two 79 yards touchdown drive, each within 2 minutes, his pass completion rate was 80% and yielding 136 Yards prior to the 4 and 2 situation (from the playbook). He look like he can moving the ball at will. I highly doubt Patriots defense could stop him.

50. Anonymous says:

And lets not forget that he HAS to score a touchdown and has 4 downs to do so. Maybe 50% in that situation

51. Anonymous says:

Someone may have already said this, but I take issue with three-down football being the norm. Why is the conventional wisdom that he "should" have punted here? It can't be because of the end result -- hindsight is obviously 20/20.

I remember learning once that human beings are peculiarly risk averse. E.g., in coin flip studies where heads wins \$100 and tails loses \$50, the majority of people choose not to play, even though their potential gains are twice as high.

In other words, maybe Trent Dilfer is just a p*ssy.

52. Dan says:

1) As others have detailed, there are so many different variables that are estimated that any estimate of GWP in this particular instance has a material margin of error, so 79% vs. 70% is by no means definitive;

2) So instead of looking at precise probability results, my take is that it was a close call. So, at the very least, the "experts" claiming that it was flat-out wrong are... flat-out wrong. Personally, I think it gave his team the best chance to win, but we'll never know;

3) Although we will never know whether it was a good move or not, people who claim that because his team lost, it didn't give them the best chance to win are foolish. The best way I could explain it to those people is that you can take a normal deck of cards and a rigged deck that has 80% red cards. If your goal is to pick a red card and you are given the option of which deck to pick from, you increase your chances of winning by picking from the rigged deck (this should make sense to almost everyone--even sports reporters and analysts). If you do so and still lose, you did not make a bad decision to pick from the rigged deck... you just got a losing outcome from a smart decision.

4) I too wonder how this will affect other coaches' (and Belichik's) propensity to "go for it." If Bill is faced with an identical situation in a playoff game in Indy in January, what do you think he would do? I'm not sure...

53. Anonymous says:

The one number that really needs to be adjusted is the 53% TD rate from the 28 yard line. That number should be significantly higher being that in a normal game setting you don't know that you need a TD therefore you don't use all 4 downs. In this situation Manning knew he needed a TD. Let's increase this number from 53 to 75%. Now we are looking at a wash...

54. Anonymous says:

"A Punt from the 28 typically nets 38 yards, starting the Colts at their own 34. Teams historically get the TD 30% of the time in that situation."

This average doesn't take into account the time left (a little less than 2 min), the number of time outs left (1), or the biggest factor of all -- a TD was needed to win the game -- which makes the defense play the game differently than, say, it would in the second quarter when the game isn't on the line

55. Anonymous says:

I love how many people read the post, take issue with a stat not accounting for something (which is a legit gripe sometimes) and then make up a number which would justify their own preconceived notion.

Hey, I think x would happen 92.6% of the time, which makes BB wrong.

The fact is you can never get the numbers perfect, but what they can show is that it was not an asinine decision, and that all of the people saying that he should have "played the percentages" have no idea what they are talking about.

I doubt BB knew the exact stats in this situation, but he probably figured something like this:
1. Our offense is damn good, and I think they have a better than average chance to pick it up.
2. Their offense is damn good, my d is gassed, I think they have at least a 50/50 chance of going 75 yards in two minutes.
3. Even if I don't make it, maybe my d steps up and makes a play, a ref makes a bad call, a received falls down. Whatever.

If that is the way a coach truly analyzes a situation (and as much as I hate BB, he is pretty damn knowledgeable), they have to go for it. To do otherwise, is to care more about what the media thinks about you then winning a game.

56. mal says:

please compute with following known data .the patriots punter Hansen when not trying to avoid a touchback had nets of 51 and 55 for an average of 53. Thus, manning starts at the 19. Add to that one time out and 110 seconds on the clock . This is the least you can do for an honest stat. Keep in mind they are playing indoors dope, kicks do go farther and higher and the colts do not have electrfying return man. The success rate based on these basic few variables is much lower for colts. And, do your stats include wining overtime drives ? You stats are alot like the junk science Al Gore is pushing on global warming-please dig a little deeper before posing as the final word with such skimpy research.

57. Anonymous says:

I think much of the debate over the exact success/failure rates misses the most interesting points from this episode:

(1) Even experts paid to make/analyze these decisions make gross errors when attempting to intuit the potential success/failure of potential strategies. I stress "intuit" becuase coaches won't have access to context depednent probabilities when making these decisions.

At the time of the play, most observers (myself included) felt that this was an obvious error, when in fact, it is most likely a toss-up at best. This means that even experts routinely overestimated the failure rate of 4th down plays (and/or underestimated the potential success rate for the opposing team after a punt). If asked, I would have guessed around a 25% success rate for 4th down plays...I would never have guessed that the success rate could be twice as high. But at least I am a lay person.

(2) That many of these same experts seem to continue to hang onto their flawed initial intuitions even when faced with more objective (even if imperfect) analysis.

I think this suggests that there are huge opportunities for teams to study the thinking on websites like this and use this to gain a small but important strategic advantage.

I actually dislike Belichick as a personality, but I think his organization may be unique in grasping concepts like this.

The one caveat I would raise is this (also raised by former players like Bruschi and Harrison). If the clear normative choice is to punt, then doing somethin non-normative is likely having some psychological impact on the Patriot's defensive players...judging from the reactions of Bruschi and Harrison, I'm not sure that the impact would be negligible or positive.

58. Anonymous says:

Your analysis seems to be flawed. Your original work concerning whether to go for it on 4th down relies on your calculation of EP. The data you used for EP excludes half of all games... You used first and third quarter data only according to your article. You specifically omitted that data in order to take out specific factors that are relevant here("I used only data from the 1st and 3rd quarters to exclude situations hurried by an expiring clock and by desperate teams or teams with large leads playing differently late in games.") How can you apply your analysis to a situation that you specifically excluded?

59. Zach says:

"dino said...

"I think a 63% conversion rate when passing on 4th & 2 is very realistic"

Brady's below 40% this year when passing on 3rd or 4th and 2."

----

Talk about small sample size. According to both ESPN and Yahoo, Brady is 9-for-14 on third or fourth and two or less (64%). I'd imagine that he's something like 2-for-5 with exactly two yards to go--please don't tell me you'd punt based solely on those five attempts, or, much less, the 14 attempts with two-or-fewer yards to go.

Great post. Not all good decisions lead to good results, and not all bad results stem from bad decisions.

By the way, I linked to your blog from my own. Take a look if you have a chance: http://kaskeyblog.blogspot.com

62. Jeff Clarke says:

"The one number that really needs to be adjusted is the 53% TD rate from the 28 yard line. That number should be significantly higher being that in a normal game setting you don't know that you need a TD therefore you don't use all 4 downs. In this situation Manning knew he needed a TD. Let's increase this number from 53 to 75%. Now we are looking at a wash..."

Ugggh!!!!

Your logic makes sense, but by the same logic, don't you need to increase the probability that the Colts score from the 65....That makes it not a wash anymore.

63. Jeff Clarke says:

There is something else I don't understand about the punt crowd. They seem to be saying that there are too many variables and too many adjustments that need to be made so the analysis is worthless. But doesn't that apply to the other side as well. What makes you think that all of the unknowns tilt towards punting? It seems like if you take the league averages and then adjust them for the fact that the offense has the advantage in both the NE Off/Ind Def and Ind Off/NE Def, that makes it more of a case to go.

I sort of view this as a poll that says Candidate A is up by 4 points with a margin of error of +/- 5 points. That means that its possible that Candidate B is really ahead by 1 point. Its also possible that Candidate A is ahead by 9. If you consider all of the intermediate outcomes between -1 and +9, A's lead is very probably positive.

I wouldn't be a 100% confident in my bet but if somebody forced me to make a bet, I'd bet on A.

It appears like the punt argument is to bet on B because you can't be sure A is ahead. But you can't be sure B is ahead either and with the evidence all pointing in that direction, shouldn't you bet on A? Remember, you must make a bet one way or the other. Abstaining is not an option.

64. Anonymous says:

I guess the biggest statistical reference would be going for it in the same exact situation. In this case. The run for Emmitt made it 0-1 and Wyche's run made it 0-2. Small sample size, but until a successful attempt of this play is made, Belichick was running a 0% chance of success (statistically speaking) with that play

65. Anonymous says:

I do not understand the source of the statistics but there is a flaw in the logic somewhere.

If the logic is correct, why ever punt?

66. gcomella@gmail.com says:

Not everything that can be measured counts and not everything that counts can be measured. Belichick should have punted but, at a minimum, when you have the best QB and WR in the league (one might argue of all time) then you have to communicate that those are the two making the play that decides the fate of the game. It's why MJ always took the last shot with the game on the line.

How can you reasonably apply backward looking data and say "the better decision would be to go for it"? Historical statistics don't account for Sunday's very unique situation in Indy. Statistics don't account for a lot of things and people (not just Belichick) may just consider that the next time someone wants to rely on the analysis.

67. Jeff Clarke says:

"If the logic is correct, why ever punt?"

Good question. Perhaps the answer has something to do with all the fans and yahoos that are saying he made the worst decision ever without actually looking at a single probability analysis. Would you rather a 60% chance at winning and a 40% chance at losing with it being someone else's fault? Or a 70% chance at winning and a 30% chance at what is happening to Belichick? Think carefully about that.

I can guarantee you if teams punted far less often, they would win far more often.

http://sportsillustrated.cnn.com/vault/article/magazine/MAG1160256/index.htm

gcomella,

Go back and read my post from 3 or 4 posts ago...

Why on earth would you bet on Candidate B?

68. mrlondon says:

Are the percentages that have been quoted here for scoring a touchdown with 2 minutes to go any different for the home team vs the visiting team? How about the chance of making a 4th and 2 for the home team vs the visiting team? And what's the percentage of a running play working vs a pass play?

69. Anonymous says:

Neat, but this is misleading. And Peter King wrapped it up well. Using baseball style averages over a long period of time and over a lot of situations doesn't work nearly so well in football where you cannot so easily play the odds, since individual circumstances and variations are much more important. Using a lot of averages based on how likely a team is to score from 28 yds or 72 or whatever, uses stats from many different teams and situations, and probably does not accurately reflect the "true" percentages (which cannot be measured of course) of THIS exact team in THIS exact situation against THIS opponent. It works on baseball because the number of situations is so large, if you can find a 55%-45% edge by doing X as a batter, then because you are at the plate hundreds of times, it can work out. In this kind of situation, it might occur once every few weeks or even every few seasons. What matters is also taking into account that the current situation might have very different %ages to the ones assumed by looking at past histories of a wide range of teams and seasons. I liked BBs call, I like "unconventional" thinking and I think it was probably a closer call than most mediots reported, however, I think it probably was the wrong call, and I am not sure I would have backed PM to drive 72 yds in that time against that defense with the clock also in play.On the other hand, I would probably have gauged Tom Bradys chance to make a 4th and 2 a lot higher, so maybe the odds worked out the same!

My point was, be careful when you use large amounts of historical data that also includes a lot of other factors and then just blindly using it on individual situations in football, compared to baseball, it is not quite so cut and dried in my opinion.

Oh, and the article trying to make out that Brady is way better than Manning. Just silly. If you swap those guys teams, Manning would have had every bit the success Brady did. the NE defense in these years has been a lot stronger than the Colts soft defense. So using direct comparisons is just silly. Of course TB has had more success in the head to heads and post season, his team was a crap load better in almost every aspect.

70. Anonymous says:

@mal: LOL at excoriating Burke's use of mixed and countering with a sample size of two. Why don't you mosey on over to NFL.com, look at Hanson's career and give us his net punting average indoors when punting from his own half. I'm pretty sure it's less than 53 yards.

71. Anonymous says:

Peter King's article was just plain amusing.

Go back and reread it. Did you notice that he started to do the math and then realized that completing it would lead to the opposite conclusion of the one he was yelling about?

Here were his stats:

The odds of making the first down: 62.5% percent (He said 60, 65 percent...I averaged).

The odds of Manning going 72 yards (weird math to get to 72 but never mind): 35%

Only one variable he left out:

The odds the Colts score from 28.

Well with algebra we can find the break even point.

.625 + (1-.625)*X = (1-0.35)

X = 93.33%

So in other words, if you used King's own probabilities you would need to believe that the Colts would score 93% of the time to make it a break-even.

Thats just to break even. If the Colts scored 92% of the time from there, Bill made the right call!! 95%...he made the wrong call, but just barely.

A professional sportswriter would have stopped and rewritten the whole article, but King had already written a long blowhardy piece talking about what an idiot he thought Belichick was before he actually did any math at all.

Once he started the math and saw it was heading in the wrong direction....well it was late, there was a deadline and he needed to get to sleep....

Read between the lines with this sort of "analysis". King might not have finished the thought but what his math (if not his hyperbolic prose) says is:

Belichick was right.

72. James says:

I think a lot of the commenters here are the same people who watch "Deal or No Deal" and form very strong opinions about which case has the million dollars in it.

Anyway, what's clearly going on here is a (mostly subconscious) backlash against the prospect of statistical analysis taking over football like it took over baseball. But here's the question I haven't seen raised anywhere else: Why is that a bad thing?

I mean, Sabermetrics (as I understand it, which is barely) means longer at-bats, more walks, fewer stolen bases, and a dozen pitching changes a game. I don't question the math, but that sucks. I want to watch a team with four guys that play like Rickey Henderson, four guys that play like Mark McGwire, and Rick "Wild Thing" Vaughn. But that's not going to happen, and baseball is less interesting as a result.

But a statistical revolution in football would be awesome. Fewer punts, more all-or-nothing 4th down attempts, fewer field goals and extra points, more 2-point conversions, more onside kicks, more going for the win instead of the tie--why would anybody be opposed to that?

Seriously. Can somebody give me a reason?

73. Anonymous says:

James,

Its an interesting question.

For one thing, I think the conventional wisdom has been really deeply ingrained. People have been watching football their whole lives. There whole lives the announcers have been telling them its wrong to go on 4th down deep in their own territory up 6. Somebody does. Its hard to grasp that guy might be right and everything they heard their whole lives might be wrong.

The other factor is a sort of nerds. Lets face it. The popular kids (ie ESPN, NBC) say to punt. A bunch of nerds (ie advNflstats, TMQ, David Romer, etc) say to go with a bunch of fairly complicated math. Very few people like math and very people truly understand it. When you say "computer model", they think a computer model is trying to outsource my job...

I honestly believe that four down football will catch on, but it will take awhile. If two years ago somebody said: Brilliant idea, I'm going to have the quarterback line up at WR and the RB line up at QB and call it the Wildcat, you'd be like "ok why?!?!". If the unconventional works it will ultimately become conventional, but it does take a little while

74. Anonymous says:

One more thing...

Deal or No Deal....

I love to watch that show and marvel at the absolutely horrendously bad math skills of the American public.

The EV at every level is easy to compute. Ok, you don't have the time to do an exact calculation, but you can estimate it very quickly.

The banker always offers considerably less than EV and then offers more with around 5 cases left.

It seems fairly obvious to me. Calculate expected value. Pick randomly. Turn down all -EV offers. Accept the first +EV offer.

Perhaps there are contestants that are doing this without saying it, but it really doesn't appear anybody is doing this. It would give me greater faith in the U.S. educational system if somebody just said "That offer is 65% of expected value. I'm going to hold out until you give me a positive expected value offer"

75. James says:

Yeah, I agree with all of that, and I agree that four-down football will catch on. I'll go as far as to predict that in the near future we'll look back on this 4th-and-2 play as the best thing to ever happen to four-down football, because it got people to start talking and, in some cases, thinking. The great irony is that it's all because they didn't get the first down (I completely disagree with everyone who's said Belichick would be hailed as a genius if the "gamble" had worked--I think the pundits would just call him gutsy and lucky and then quickly forget about it).

It just makes it so much more baffling that there's an option that's (1) strategically preferable and (2) more fun to watch, and people are so adamantly opposed. Does everyone hate fun?

As for Deal or No Deal, that raises an interesting point. I agree that the contestants aren't exactly mathematicians (for the record, neither am I), but I wouldn't go strictly by EV. If I had a choice between \$300,000 and a 50/50 shot at \$0 or \$1,000,000, I'd take the 300K. There's all sorts of stuff I could do with 300K that I couldn't do with 0 (i.e. that I can't do now), but the difference between 300K and a million doesn't nearly as important.

I think that's a rational way to think when you're dealing with your own livelihood, but people use the same thought process with football, and that's entirely irrational. The obvious differences are, in football (1) there's no guaranteed middle option, it's either win or lose (ignoring ties, because ties in the NFL are ridiculous), and (2) there's no serious consequence for losing (you might miss the playoffs, and your coach might get fired, but nobody's going to get hurt or wind up homeless because of a loss).

76. Anonymous says:

Isn't there something missing here.

BB goes for it.Six times out of ten he comes out the other side of the 2 minute warning as the most likely game winner.(WP around 100%).But 4 times he comes out as the most likely loser because Indy stop them.

If he punts,he's almost certain to still be the most likely winner after the punt,although his win probability will be reduced compared to the best and average case scenario of going for it.

Given that he only gets to play this play once,shouldn't he trade off part of his longterm win probability (which he only gets if he continually goes for it on 4th down)for the knowledge that he will still be the most likely winner if he punts the ball away.

77. Anonymous says:

With all due respect, I think your post illustrates part of the disconnect between the math people on this site and the conventional wisdom.

The conventional wisdom relies heavily on rounding.

Punt = 70% Win Prob (WP) = probable win

Go for it and make it = 99% WP = probable win

Go for it and miss = 43% WP = probable loss

Go for it and make it and all you've done is turn a probable win into a probable win. No real difference. Go for it and miss and you've turned a probable win into a probable loss.

I guess the problem is people are too quick to round. Yes, a 70% WP means that you are likely to win but describing a 70% probability as "almost certain to still be the most likely winner" is incorrect.

Alex Rodriguez has a lifetime batting average of .305. That means (disregarding walks and sacrifices) he is out 70% of the time he comes up to bat. With A-Rod at bat, would you say he is "almost certain to win"?

People seem to treat 50% as some sort of magic number. Slightly above it and you are the favorite. Just cruise and you'll win. Slightly below it and you are the dog. You're in a lot of trouble. How did you get here exactly...

I guess the point is that 50% isn't a magic number. You can't just round up or round down. If you are risking turning a slight favorite into a slight dog but in return you get a 60% chance to turn that favorite into a virtual guarantee, you should take it.

78. Anonymous says:

"Given that he only gets to play this play once,shouldn't he trade off part of his longterm win probability (which he only gets if he continually goes for it on 4th down)"

Another misperception....

A casino invites you to be a blackjack dealer and keep the profits. Or you can sit on the regular side play like a regular gambler.

Play 10,000 hands from the dealers side and you will almost certainly be profitable.

Play 10,000 hands from the players' side and you will almost certainly lose.

But if you play one only hand, you will still have better odds from the dealer's side. Yes, there is variance and you might win or you might lose from either side, but the smart move is to sit in the dealer's chair even if its only for one hand....

79. Anonymous says:

I've read through most of the arguments and finally the last anonymous poster is the only one here who seems to get it. It a risk-reward type call and I think the call was way too risky in this situation.

As a coach myself, I like the idea of creating the most opportunities for my team to win. By punting you force the opposing team to beat you and they have zero room for error. Your defense will have multiple opportunities to force an error or make a stop while having time on their side.

So IMHO, by taking this huge risk and going for it, you are in fact showing zero confidence in your defense.

80. Anonymous says:

Anony 1.
"Yes a 70%WP means that you are likely to win but describing a 70% probability as "almost certain to still be the most likely winner" is incorrect".

You're miss reading me.I'm saying if you punt in the vaste majority of those punts you come out of it with a higher WP than Indy.I never said anything remotely resembling "a 70% chance means you are almost certain to win."

Anony 2

"But if you play one only hand,you still have better odds from the dealers's side."

So if we toss a fair coin once and if you're corrct I give you a million pounds,but if you're wrong I take your house,you're playing are you?.....Or is the downside on this favourable bet too risky?

D.

81. Anonymous says:

He (and the team and his other coaches) only had 25 seconds to not only decide YES or NO but to a play... against arguably the BEST team in the NFL. They didn't have 24 hours and computers and replays and historical games to look back on... or a bunch of 'Monday Morning Quarterbacks' or 'Yahoos'. GREAT CALL ! Awesome Call. Gave his Offense AND his Defense a chance. Colts had to score a TOUCHDOWN ! Not 'just a FG'. We missed a few other chances to add more points earlier. Oh... and by the way... how come non of the whining yahoos are complaining about the Colts running up the score on the Pats by scoring 35 points and not letting up? I guess the Pats should have taken it easy on the Colts earlier and only scored 34. Oh wait... they did. Great Game. 198,000 people were there. I know I was one. Ha ha ha.

82. Anonymous says:

Going for it in that situation is too risky because if you fail:

1) The game clock becomes a non-factor.
2) The offensive play book is wide open.

All these statistics are great but the fact is Indy had only scored on 4/13 possesions in the game and 2/5 in the second half. So what are the real odds of them scoring again with only 2 minutes left and 1 TO?

He had the big stack and went all-in with a marginal hand and when he lost he wasn't out of the game but he was crippled.

I say keep the advantage and force the other guy to play. Obviously he felt Indy held the advantage and, based on his decision, he had zero confidence in his defense.

83. Anonymous says:

Please stop the madness. Enough of the defense of the mad scientist. There's no way to defend the call; it lost the game for the Patriots. The Pats website says it all - 2009 3rd down conversions 56/123 (45%); 4th down conversions 5/11 (45%), so converting (in 2009) is worse than a coin flip. If Belichick is so worried about giving the ball to Manning in this game, why would he decide, more likely than not, to give it to him with a short field? Methinks many statisticians overlooked the fact that the mad-one had such great success with unconventional 4th down conversions in his early season game at Atlanta that he forgot one key fact - probabilities are just that, probabilities - the outcomes are binary - success, or in this case, failure.

84. Anonymous says:

I agree that going for it on 4-2, even in that position, is the true mathematical favorite. However, in practice I only agree with the decision if the defense is told to allow a touchdown if/when the Colts are very close... tackling the runner at the 1 with 1:20 left (and no timeouts) ultimately is what cost the Pats the game. With a minute and only needing a field goal Brady is successful most of the time (say 75% but I am sure someone can correct).

85. Anonymous says:

Probabilities are very interesting to consider (this is the Tony LaRussa format of coaching), but I think Belichick did the RIGHT thing, considering his offensive personnel vs. his defensive personnel, and how deadly Peyton Manning is in a must-win situation. The problem was the play execution on 4th and 2. Why have a guy go to exactly 2 yards to attempt the catch (which he made, but juggled so fell back to 1/2 yard shy)? If Faulk (who had a GREAT game) would have gone out 1-2 yards further, the 1st down would have been made and all the second-guessing would be history . . . not to mention Maroney's ridiculous fumble at the goal line, which cost the Pats one score, and Brady's interception in the end zone that likely cost them at least a FG try. The Pats had a great offensive game (except for the 2 miscues I just mentioned), but this defense (though good at times in the game) is not really a Super Bowl defense. I think the Colts are quite overrated, and will not be your Super Bowl champs, though I don't think NE will be either. It was a devastating defeat, but considering how weak the AFC East is, it will not cost them the division title - only best record chances for the playoffs. The boring coaches (that DOMINATE the No Fun League) always opt for easy choices like punting. I liked the gamble, and am sorry they did not execute the 4th down and 2 play better, but it should have been successful.

86. Anonymous says:

I think this is not nearly as complicated as people are making it. Belichick normally punts on 4th down. He would have punted here but he did not think Manning could be stopped by his D. He had to quickly decide if he wanted Tom Brady deciding the game or his D. Since Manning was the other QB he probably looked at a punt as giving him a 30% chance to win and going for it is 50/50 at worse. Any other QB in the league and he punts.

87. Anonymous says:

Here's why Belichek's choice was too risky...

NE has ~70 yards to go, same distance to endzone for IND if NE decides to punt.

Therefore, we can assume NE's WP is at least 70% BEFORE the decision, but not much more because they're only up by 6 with 2 mins. to go.

AFTER A PUNT, NE's WP stays about the same at 70% (assumes 40 yds net to IND 30)

4&2 Make it: NE WP% jumps to 99%
4&2 Blow it: NE WP% falls to under 50%

REDICULOUS! He could have kept his WP about the same and force IND to go the ditance 3 times in row!!

88. Anonymous says:

Applying .60 conversion factor is way off. NFL 4th down conversion in 2009 (3 yards or less) is .449. The Pats are 5 for 11 this year or .455 which is only average. When the conversion attempt failed did anyone actually think the Colts wouldn't score?

Even conservatively, if the Colts/Manning had a 70% chance to score (if 4th down conversion failed), and based on NFL 4th down conversion stats for 2009, the Pats would have a WP of only 62%, going for it. From this website, the league average is 27%, scoring from own 28 (more realistic punt based on Pats average that day)- that would be a WP of 73%, for punting. Manning is great but is there close to 40% chance the Colts would drive 72 yards for a TD in less than 2 minutes? Possible, but a tall order. Looks like Belichick should have called for a punt, but apparently didn't have any confidence in the D, or just plain screwed up.

89. Anonymous says:

Does anyone consider that the decision was a win-win for Bill (despite what the media coverage says)?

A) Pats make 4th down, essentially the game is won.

B) Pats fail to make 4th down, defense holds Colts offense ... Pats win.

C) Pats fail to make 4th down, defense allows Colts to score quickly. Pats offense roars back with game-winning drive of its own.

And finally the winning situation that BB has created ...
D) Pats fail to make 4th down, Colts score to win game - no-one gives them credit for the victory. Colts players don't actually know whether they could have driven 75 yards in the final 2 minutes against the Pats defense. A spectre of doubt looms as to whether they could have done it plus Pats defense gives nothing away about how it would defend that situation. They're still in prime position of the AFC East to make the playoffs.

The questions for me are:
- would Bill have made that same choice in the Super Bowl? (As opposed to a regular season game where he has 7 more games to qualify for the playoffs).
- would he have made that choice with 45 seconds left in the game? (Over 1 minute - same decision; 30 seconds - punt; 45 seconds much more in the balance as to whether Peyton can drive the Colts 70 yards in that time).
- if the Colts had been held on that final drive and faced a 4th down at say the 20 with 1 min left; would they have gone for the 4th down OR would they have kicked the FG hoping to recover the onside kick?
- when is Bill going to do something about the Pats defense tiring late in games? (See 2006 AFC Championship, Super Bowl XLII).

90. Anonymous says:

I didn't see a comment discussing how Manning and the Colts do in the final two minutes, when they are tied or behind. Surely there's a stat for that floating around somewhere?

91. spoonfulofpeter says:

Also, as Tony Dungy mentioned on SNF, 60% is a league wide average of 4th and 2 situations. That includes a large number of situations where teams are down multiple scores and facing prevent defenses. This skews your entire calculation as their conversion rate has to be way less than 60%. The Colts were not playing prevent. They were playing do or die.

92. James says:

spoonfulofpeter,

I'm pretty sure Brian's numbers include data from only the first and third quarters. That ensures the results aren't skewed by prevent defenses or desperation come-from-behind situations.

93. Anonymous says:

Morons, who don't know how to calculate risk.. And there's really no point in trying, with morons like you guys..

94. Anonymous says:

I realize i'm pretty late on this...but where everyone (including this site) is screwing up is giving the pats a 60% chance of picking up that 4th & 2. that's the chance of picking up a normal 4th & 2 where it's not the worse thing if the other team doesn't stop them because they're still concerned with the deep threat. there was no deep threat on that play and was closer to a 2 pt conversion...which has a far less chance of being successful. that's where belicheck screwed up with his stats

95. Anonymous says:

I love the Colts but this really surprised me.

Rick

96. Anonymous says:

The truth of the matter is that play-calls are judged by their outcome. If Belichick's decision ended in a first down, it would have been the correct decision. But the Pats did not make the first down. The decision was a bad one. If you ask any coach how they know if they made a good play-call or not, their response will, most likely, be "if it works."

Furthermore, the statistics for this analysis are general. Historical teams were not playing that day. The average quarterback was not playing that day; Peyton Manning was playing that day. One would assume that he would have a better chance of scoring from both the colts 34 and the Pats 28. His statistics are the only useful ones. Historical statistics from the whole leage are not viable in this situation.

Also, the statistician did not take into account the possibility for a negative play, interception, sack fumble, defensive touchdown. There are many more variables than just make it, miss it, or punt it. Also, in both situations (punt or miss the attempt) the Colts had two possibilities of winning the game: a touchdown or a fieldgoal, onside kick and field goal or touchdown. The latter would be more probable from the 28 than from the Colts 34 also. These statistics you give are essentially useless because they do not take into account the multiple amount of variables that exist in the game. The variable are almost limitless. Lets try to list them.
Punt: chance of blocked punt, chance of punt return for touchdown, percent of muffed punt, average net yardage (you did this), PEYTON MANNING'S chance of scoring from his 34, Chance to kick FG recover punt and score again, etc.
going for 4th down: chance of making it, chance of sack-fumble, chance of sack that would affect field position, chance of interception that would affect field position, chance of defensive touchdown, chance of PEYTON MANNING scoring from Pats 24, chance of FG then onside kick and score.
All are possibilities and must be taken into affect of the decision, and none even get into the probability of that play-call being successful. percent of balls caught by Faulk in contrast to the number he is targeted (the play-call was only designed to throw to Faulk). Percentage of 4th down attempts being successful for the offense, but ALSO percent of 4th down attempts being successful against the Colts. The decision made by Bellichick is far more complicated than chanc of scoring after punt vs percent scoring if you don't make the first down multiplied by the percent chance scoring from the Pats 24.

97. Anonymous says:

Your stats are completely irrelevant. Every team lines up facing a different situation.

98. Anonymous says:

those "stats" are complete bs! do they say if half the other teams players were hurt before that play and subs were in? or how bad the oppositiuons defense and yours was great? or etc, etc.? how about you give me one time when a team up 6 points with about 2 minutes left in game which should have been less as you should have forced the colts to use up their remaining timeouts by not throwing an incompletion then wasting a timeout yourself taking surprise out of it as colts knew you were trying on 4th down now ? btw - manning had been intercepted in the 4th quarter. so tell what happened the previous time in a near similar situation the percentages when it happened! YOU CAN'T! Because there were no similar situations ever so those "stats and probabilities" are completely irrelevant!

99. Anonymous says:

those "stats" are complete bs! do they say if half the other teams players were hurt before that play and subs were in? or how bad the oppositiuons defense and yours was great? or etc, etc.? how about you give me one time when a team up 6 points with about 2 minutes left in game which should have been less as you should have forced the colts to use up their remaining timeouts by not throwing an incompletion then wasting a timeout yourself taking surprise out of it as colts knew you were trying on 4th down now ? btw - manning had been intercepted in the 4th quarter. so tell what happened the previous time in a near similar situation the percentages when it happened! YOU CAN'T! Because there were no similar situations ever so those "stats and probabilities" are completely irrelevant!

100. Justin Becker says:

Great blog, I have book marked and plan to share with my friends later :) Yahoo Fantasy Football Rankings