Advertisement

Belichick Made Powerful Enemies of the Analytics Crowd by Going for the Field Goal on 4th & 3

To paraphrase Shakespeare, "Uneasy lies the head that wears a headset." When you've coached as many games as Bill Belichick and precious few of them have been meaningless, you make a shit ton of decisions that have far-ranging, history making implications in the heat of the moment. So many that even the most casual observer can rattle them off without wracking their brain. 

Going for it in the Super Bowl against the Rams instead of playing for overtime. Throwing touchdowns to Mike Vrabel in multiple Super Bowls. Snapping the ball out of the end zone in Denver for the intentional safety. Kicking the ball away to Peyton Manning in overtime. The playoff game against the Ravens that included not only a Julian Edelman-to-Danny Amendola double pass, but all those Ineligible Receiver alignments. Not calling time out with the Seahawks inside the 5 and time running out. Letting Stephon Gostkowski try a 41-yarder instead of going for it on 4th & 2 in Super Bowl LIII. 

Then there are the ones that didn't work out. The Super Bowl That Shall Not Be Named I, when he went for it on 4th & 13 instead of letting Gostkowski try a 48 yarder. The instant legend that was 4th & 2 against the Colts. The fake punt in the playoff game against the Jets. Of course, benching Malcolm Butler and not unbenching him.  

And now some are including his call to let Nick Folk attempt a 56-yard go-ahead field goal in a rainstorm, rather than give Mac Jones the chance to convert on 4th & 3. It felt to me that it was a tough call either way. Though the sort of tough call that one should trust no one more in the history of the human race to make. But not obvious either way. 

Unless of course, you're big into analytics and familiar with the computer models that produce the Win Probability stats. Whether these are the sort of software used for Season Mode on Madden, or algorithms like the ones that keep my YouTube page loaded with Norm MacDonald content, or more like the USS Enterprise's computer with the clumsy robot voice, your guess is as good as mine. I just know that NFL's NextGen stats begs to differ with the highly advanced organic computer operating underneath Belichick's rain-soaked hood:

Source - This decision by Belichick -- which cost the Patriots 10.3 percentage points in win-probability value -- is one of the more complex calls we've seen this season, with so many different factors influencing the final numbers. Let's break down each of the components:

Conversion probability: 

  • Folk had a 45.1 percent chance of making a 56-yard field goal, according to our field-goal probability model, which takes into account real-time weather and humidity data.
  • If the Patriots had gone for it -- with 4.4 yards to go, as measured by ball-tracking data -- their chances of converting were 52.3 percent.

Win probability:

  • If Folk had made the kick, the Patriots' win probability would have been 52.3 percent -- given that Brady and the Bucs would have gotten the ball back with just under one minute left, with two timeouts and trailing only 2 points, this was almost a 50-50 proposition for the home team.
  • If Mac Jones and the offense had converted for a first down, their win probability would have jumped to 65.3 percent, with Folk presumably having the chance to end the game with a more probable field goal, with less time on the clock for Brady and Co.
  • Missing the field goal or failing to convert would mean a loss, as we saw.

By combining the likelihood of each outcome with its effect on the outcome of the game, we can calculate the expected win probability of each decision. Before the play, the value of kicking a field goal was 24.3 percent. The value of going for it was 34.7 percent. The difference -- 10.3 percent in favor of going for it -- represents the magnitude of the recommendation.

Let me stop here and say I'm not taking the low-hanging fruit of just dismissing this as the work of nerds and make some obvious joke about them getting stuffed into lockers. I enjoy this sort of analysis to a degree. I think any organization that is serious about doing whatever it takes to succeed does as well. 

And I know for a fact the Patriots take these kinds of models into some consideration, particularly in situational football ... um, situations. (That sentence got away from me but this is no time for the Thesaurus.) I wish I could remember the names and places of this story, but I remember years ago some professor of economics on the West Coast wrote a paper arguing that coaches at all levels should go for it on 4th down, every time, regardless of the situation. A few coaches around the league were asked about it and laughed it off with the old, "When I start listening to ___, I'll end up working as a ____" line. But when Belichick was asked, he said he'd read the paper, found it interesting, agreed with some of the conclusions, but ultimately felt it didn't take into account the human element. How the emotion can come out of a team if its offense fails to convert deep in their own end of the field and so on. The author of the piece was asked about Belichick's comments and was gobsmacked that an NFL head coach with multiple rings had read a paper he'd written 3,000 miles away, especially since it had never been published

But therein lies the problem with NextGen's analysis. They can input all the data, take into consideration all the variables based on past outcomes of countless such situation and come up with suitable mathematical results. I studied some of this stuff in a business class called Quantitative Analysis and know it's how they do everything from knowing how much to supply a deployed army to where to dig oil wells. So it's not bullshit. But what it's missing is the human element. 

The reason it made sense to go for it on 4th & 2 against Indy 12 years ago was a human element. The element being one specific human, Kevin Faulk. When Faulk went into the Patriots Hall in 2016, Belichick referenced how automatic he was when it came to converting those high leverage, short yardage plays. 

So many of his plays were just, third-and-six and he got seven, third-and-four and he got five, third-and-three and he got four. He just had a great knack [for making plays], like Troy [Brown] did. [He was] a very instinctive player; had a great knack for playing the game. He always seemed to do the right thing, even if it wasn’t a play made, maybe there was no more than what he could get, he got what he could get. He did the right thing, he made the right play.

In fact, not only do I still think Faulk picked up the yards he needed on 4th & 2, I said at the time that instead of bringing out the sticks and measuring, the officials should've just looked at the 33 on his jersey and signaled, "1st down!!!" 

The issue with the Patriots on October 3, 2021 is that neither Faulk nor New Faulk, James White were on the field. White was lost for the season a week earlier and as of the moment, they lack that sure thing, can't miss, go to, gotta have it guy. They may have one on the roster right now, but no one knows with enough certainty who that is at the moment. So they put the ball in the feet of their most reliable player over the last couple of years. One who said after the game he was connecting on kicks of 58 yards. 

Advertisement

Factor all that into your equations and an already narrow range of probability - 45.1% to 52.5% with an almost 50/50 chance Brady drives the Bucs downfield for the win anyway - and the decision to kick makes enough sense that it's not worth arguing to the contrary. These games almost always come down to a half dozen plays or less. This game was lost on at least that many. Missed stops on 3rd & long, a fumble, failure to pick up blitzers. So the fact a kicked ball missed its mark by half a foot should come as no surprise. 

If you really want to feel like something cost them the game, focus instead on the play before the missed field goal. 

That is this team's margin for error as they start to learn how to win. Dammit all.