I love you Tough Tom, but you were wrong on 1st and goal

After the Giants got their final first down last night, setting up first and goal, I tweeted the following:

Twenty years ago, neither team manages the clock correctly in this situation.

I was both right and wrong. The situation was the perfect moment to put the modern advance in sports strategy on display. There was just enough time left to almost run out the clock, given the two Patriots’ timeout remaining. The Giants were down exactly two points — down three and they’re trying to score a TD, down one and they’re arguably trying to score 8 to get a full touchdown lead. And the defense was being coached by a man who was almost certainly aware of the strategic implications. That it happened in the Super Bowl was almost too good to be true. If both teams had played 1st down perfectly, it would have become the poster-child for the advancement of strategic thinking in football.

As it turns out, the Patriots did play it (almost) perfectly — they opted to let the Giants score on second down, so they could get the ball back (they probably should have done that on first down, to preserve their other timeout). The Giants, on the other hand, arguably made two mistakes. One was Bradshaw not being able to stop himself on the 1-yard line. The other was the decision to run the ball at all; they could have just knelt on it three times at the seven at kicked a 24 yard field goal with very little time on the clock.

But let me be crystal clear here: with 1:09 left, 1st and goal, down two, opposition with 2 timeouts, it is absolutely correct strategy NOT to score a touchdown. Whether you try to get closer than the 7-yard line is debatable — it depends on the relative success rate of 24 yard field goals vs. 18 yard field goals, as well as the probability of fumbling in both scenarios — but you certainly don’t want to get in the endzone.  The theoretical math is very easy: if you kneel three times, you will be able to kick a field goal with, at most, about 20 seconds left on the clock (and probably less, since the kneel-downs take more time than you think, and can be prolonged). Even if you do run a real play, you absolutely stop on the 1 yard line and do the same thing. If you score the touchdown on first down — which you will if your opponents are correctly letting you score —  the opposition will be down either 4 or 5 with about a minute to go, with two timeouts. So the question is simple: which is more probable — missing what amounts to an extra point, or Tom Brady leading a touchdown drive in 1 minute with two timeouts?

It’s not even close. You kneel and kick. Or stop on the 1 yard line.  League wide, 99.4% of extra points were made this year. The Giants were 45 for 45. You think Brady has less than a 0.6% chance of leading a TD drive with a minute and two timeouts? Not a chance. According to the NFL win probability stat, the Pats had a 4% chance to win when they got the ball back. And they only had 1 timeout as it turned out. And win probability doesn’t take into account the individual team, or whether or not you have Tom Brady. Here’s the thing: football is a zero-sum game. If Belichek was correct to let the Giants score, then by definition the Giants were wrong to get into the end zone there. And vice-versa. By the above math, the Giants gave the Pats  roughly 24-1 odds to win, when they could have made it roughly a 199-1 chance. That’s right: by getting in the end zone, the Giants increased their chance of losing roughly eightfold. (This math doesn’t include what the Pats could do with 10-15 seconds and no timeouts, down 1, after your field goal and the ensuing kickoff. But that’s virtually negligible, especially without a timeout to get the kicker on. They are basically reduced to a hail mary from their own 20. If you want to give them a 1% chance of winning that way, go ahead, it doesn’t change the strategy).

After the game, Coughlin admitted he didn’t send in the order for Bradsahw to try to stop on the 1-yard line — that was Eli. Instead, Coughlin said he actually wanted the touchdown there, arguing that no kick is ever guaranteed. Bradsahw said basically the same thing. That’s almost the perfect expression of risk-averse coaching, which is a huge problem in the NFL. If you deviate from conventional wisdom — no matter how much it increases your probability of winning — and it doesn’t work out, you get killed by the media and popular opinion. Consequently, coaches have incentives not to maximize their chances of winning, if doing so has them implementing unconventional strategies that will be criticized (nad possibly get them fired) if they fail. If the Giants had somehow blown a field goal and lost the Super Bowl after Bradshaw knelt on the one, Coughlin would have been (incorrectly) buried by the media. But if they score the TD and then the Pats come back and win, that would have been (wrongly) seen as not quite so bad: we did everything we could, but they beat us.

It’s horrible logic, but it still holds in popular perception. Let me repeat: the Giants hurt their chances of winning last night by scoring that TD; the actual outcome doesn’t matter when evaluating the strategic decision. But perhaps the conventional wisdom about these things will not hold for much longer. Last night definitely exemplified how dangerous the traditional decision can be, and there’s actually a debate going on today in the popular press that leans toward kneeling and kicking.  Twenty years ago, that debate would not even have existed; no one would have dared even consider not getting the TD.

Share

26 thoughts on “I love you Tough Tom, but you were wrong on 1st and goal

  1. John

    Totally agreed. At the two minute warning the Giants had the ball on the 18 yard line, so that’s a 35 yard field goal, which is probably good, what, 95% of the time? Belichick himself wouldn’t have let them walk in even 5 years ago, that’s how rapidly thinking has changed on this issue (I can think of a lot of games where they should have let the opposing team score but tried to stop them and instead ran out of time when they finally got the ball back).

    Unrelated, very odd that 12 men on the field does not mean time is put back on the clock. What is preventing a team from putting 18 guys on the field and taking the 5 yard penalty? Apparently Buddy Ryan had a defensive formation that called for 12 guys to be on the field for end-of-game situations.

    Reply
    1. Matt Post author

      Mike and Mike brought up the 12-man thing today. Golic said Buddy Ryan once had them put 15 guys on the field in a similar situation, to just take the 5-yarder each time. Definitely bad rule. M&M suggested some changes — put the time back on, making a second 12-man penalty in the last 2 minutes a 15 yarder, etc. But it definitely should change.

      m

      Reply
    2. B-Rob

      You kill the clock with the penalty, which is why you do not want to intentionally put 12 on the field. In this case, had Brady completed that pass but not had the penalty, the clock probably would have run way down before the Pats got lined up. But with the penalty, they clock would have stopped, permitting the Pats 9 seconds and one play from about the three.

      Reply
  2. Timbo

    On paper, this argument makes sense. But in the actual game, I believe the Giants made the right call. There is no such thing as a “routine” Super-Bowl-winning field goal. Just ask Scott Norwood. Take the TD, force the Pats to score a TD themselves, hang on, and win.

    Reply
    1. jeffs12a

      Agreed. Matt’s argument misses one statistical issue: Just before they scored the touchdown, the Giants were losing the game. In order to force the Pats into a game-winning drive, they had to go ahead and take the lead. The real question is this: Do you stand a better chance of making the field goal from 1 yard out? Or do you stand a better chance of walking into the endzone undefended??

      Reply
  3. Jay

    It’s tough to tell a running back not to score when he’s got nothing between him and the white line but grass. It goes against every instinct in his body to know lower his shoulder and get every last inch until he gets the score. Frankly, I’m surprised Bradshaw came as close as he did to taking a knee.

    Reply
    1. matt bianco

      they had 3 timeouts when the challenge was made, leaving them with 2. i dont remember the other timeout called to give them 1 left on their final drive. challenging with 2 timeouts left would have been ill-advised, i do agree. but from my memory they had 3 at the time.

      Reply
      1. Matt Post author

        They called timeout after first and goal — remember, they didn’t let the Giants score until second and goal. That was certainly a mistake; if you are going to let them score, doing it on first down and saving a timeout is optimal. But yeah, they spent one timeout on the challenge, the second stopping the clock with 1:04 to go (after first and goal), and the third after they got the ball back.

        Matt

        Reply
  4. Pingback: Why the Patriots were right to let the Giants’ Ahmad Bradshaw score a touchdown | Saint Petersblog

  5. Dan MacDonald

    Matt –
    Interesting article, but your math is faulty. If the Giants kneel and kick a field goal, as you say, the chance of the Patriots’ scoring is not 4%, because they don’t have to score a touchdown anymore. It is probably much higher because all they need to win is a field goal.
    Obviously, the two variables here are time and points. The perfect clock management situation for the Giants would have been to run the clock down (kneel? Run short distances and then fall down?) and then go for the touchdown. After burning some clock, if you can’t get into the end zone, then you kick a field goal. But to say that you don’t want to score a touchdown is nonsense. You have to score one way or the other. You might as make the hurdle as high as possible once the Patriots get the ball back.
    The fault on Coughlin is not that he went for six; it’s that he didn’t burn more clock before putting the ball in Tom Brady’s hands.

    Reply
    1. Matt Post author

      Thanks for the comment. I think you are confusing the issues. It’s true that what happens after the Giants kick the FG matters, but the probability of the Pats scoring at that point is not higher tha 4%, it’s massively lowerand utterly negligible to the calculation.
      First, the Giants may have been able to kick the winning field goal with no time on the clock, such that the Pats never even got the ball back. In any case, the Giants would have been able to use up the Pats timeouts and get the clock down to no more than 17 second, even if you don’t count the time the kneel takes on 3rd and the time the kick takes, which could easily be 5-7 seconds total. Add on the time the kickoff takes too. Worst case, after a touchback, the Pats would have had about 13 seconds and no timeouts, on their own 20. It’s amost impossible to turn that into a FG – less than 1%chance at any rate. Bottom line is that the Pats would have been playing for a TD there.

      Also the 4% figure is for winning down 4 with one minute and one timeout. The equivalent figure for 15 seconds, no timeouts, down 1 is much, much lower.

      In fact, its so low you can discount it. But for the sake of argument, call it 1%. That means kicking gices the Pats a 1.5% chance, while scoring the TD gives them a 4% chance. It’s still a no-brainer to kick.

      Reply
  6. Pingback: The Super Bowl and the Election « David Klion

  7. Ken

    Since when is 99.4% a sure thing? The Ravens missed a short field goal a couple weeks back…and the World Series and the confluence of events on the final night of the regular season in baseball proved that as much as 99% odds sound like a lock in a given situation too many things can go wrong. And do we remember the short field goal attempt the Cowboys botched in 2007 against the Seahawks?

    This reminds me of the last debate over how much of a genius Belichick is…the game against the Colts with the failed fourth down attempt deep in their own territory. Sorry…in both instances he basically allowed the other team to take the lead…and last time I checked giving up a lead is the first step to losing a game. That isn’t genius. It’s stupid.

    Reply
    1. Matt Post author

      It’s not a sure thing. It’s just more of a sure thing than taking the TD. If you needed to make 2 free throws to win a baketball game, do you want the 99.4% guy on the line, or the 96% guy? That’s the question.

      Matt

      Reply
      1. Ken

        I would rather have the lead with the pressure being on the other team. A team that had not completed a pass for more than 21 yards all game. A team whose wide receivers and tight ends were basically shut down for the final 20 minutes of the game. A team that would have to move 80 yards with one time out and less that 55 seconds on the clock.

        A 100 percent chance of a four point lead versus a 99.8 percent chance of winning? Sorry, fresh memories of the Ravens two weeks ago and deeper memories of Cowboys vs. Seahawks in 2007 have me willing to take the risk of seeing Brady and company do something they haven’t done since 2004.

        Reply
  8. B-Rob

    Wrong wrong wrong. Just wrong. For one thing, while it is true that extra points have a high conversion percentage, I would be willing to bet that if you look at short field goal percentages in the Super Bowl (or even just the playoffs) it is lower than that. Think “Romo . . . Seattle . . . .”

    Second, regardless of the circumstances, I would rather be up four kicking off with a minute left than down two with six seconds left. You are snatching victory from the jaws of defeat by trying to play the clock and not the scoreboard. Indeed, ask John Harbaugh how that works out for ya!

    Third, for whatever reason, Tynes was not sharp yesterday. He was pulling everything left, which could be a real problem with the angle you have on a short field goal.

    Fourth, the Giants D had handled the Pats. Indeed, but for a wonderful play on 4th and 16 (picking up 19), they had the game nearly won immediately after giving Brady the ball.

    Reply
    1. Stan

      Totally agree. Bad snaps, bad holds, blocks, bungling timeouts, misses by kickers under pressure … we have seen so many ways that last second field goals can go wrong. Not hypothetically, but in actual play. Bird in hand theory wins here for sure.

      Reply
      1. Dan C.

        I completely agree with Matt’s assessment. I was screaming at the tv for the Giants to kneel, and if the Pats had scored on their last drive, I would have been sick for the rest of my life.

        Bad snaps… botched holds… pulled kicks… anything could happen. Right. There is a downside to everything. INCLUDING GIVING TOM BRADY THE BALL BACK!!! What about Tom Brady marching down the field and scoring a touchdown with 57 seconds and one timeout? That happens – oh – one in five? One in four? One in three? Especially against the Giant defense, which has been susceptible to no-huddle and hurry-up drives all year. Look at the Dallas game in Dallas – before JPP blocked the potential game-tying FG, Romo drove Dallas 50 yards down the field in 46 seconds, with zero timeouts. How about last year against Philly, when Vick shredded the Giants defense for a billion fourth quarter points en route to the most amazing comeback in NFL history? Or the regular season Packer game this year, when Eli’s game-tying TD pass with :58 seconds left proved insufficient to send the game to OT. Rodgers drove the Pack 70 yards on four plays to set up the game-winning chip shot – and they had time left to try for a touchdown, but they didn’t need one.

        On Sunday, the Giants barely made it out of the last drive alive. The ball to Branch on 1st down was barely tipped by Phillips and still could have been caught; Branch had another 20 yards in front of him at least; it would have been a forty yard gain. But forget that play – what do you think is more likely: Tynes misses a 25-yard FG, or the Pats pull down the Hail Mary? Which of those two things – on their own – is more likely to happen? I think the Hail Mary is more likely to be caught. I mean, it almost was. The Giants weren’t covering the trailing receivers! They had two CBs standing off to the side, watching the ball get tipped by Hernandez, while Welker and Gronk trailed the play like they’re supposed to do. Given Tynes – 45-for-45 on XP this year, and 12-for-12 on FG under 30 yards. He was 12-for-12 on FGs under 30 yards in 2010, too, and hasn’t missed an XP since 2008. That’s a whole hell of a lot of snaps the Giants have managed not to botch. In fact, since Lawrence Tynes last missed a kick from inside 30 yards, Tom Brady has engineered 6 4th-quarter game-winning drives. And, in that span, Tynes has had four times as many kicks from inside 30 yards than Brady has had 4th quarters.

        Had the Giants taken three knees on the 7 yard line and kicked a FG, they would have kicked-off with about 15 seconds left. The Pats would have had to have tried to return the kick-off for a touchdown (which they didn’t do all year). After that failed, they could have tried some sort of hook and lateral nonsense… they had no chance. It was down to a chip shot field goal. Tynes hasn’t missed one in years. Is it possible for him to miss one now? Of course. Is it more probable than Brady leading the Pats down the field for a touchdown with :57 and one timeout. Not a chance in hell.

        The Pats faked their way through that Super Bowl with dink-and-dunk, screens, traps, reverses – anything they could think of to mask the fact that they were physically overmatched from the starting gun. They wiled their way to a 4th-quarter lead, and they almost outsmarted wiled their way to a win in the endgame. The logic Matt outlined above would be coming out of every set of TV flapping gums from Skip Bayless to Katie Couric. Of course, I couldn’t have chimed in. Because if the Giants had lost that game in THAT fashion, I’d still be throwing up.

        Reply
  9. Michael

    Citing the extra point conversion percentage is misleading, because the vast majority of extra points are not aggressively/all-out attempted to be blocked. In this case, an all-out block will be in order. Factoring in that, as well as the small probability the pats still score with the 15 seconds left via kick return/otherwise, makes it closer.

    Still agree with what you’re saying though. Sometimes people fail to realize that the scoreboard can sometimes be distracting when considering one’s odds to win the game. The goal should be to maximize the odds of winning, period, even when it might mean only leading for the final second.

    Reply
    1. Matt Post author

      True, but the win probability for one minute left and one timeout is also misleading, because Brady is much better than the average QB. So things like that are generally a wash when you are theorizing — the individual situation might vary slightly from the stylized data at hand, but the general conclusion still holds.

      But I take your point.

      Reply
      1. Michael

        Agreed that Brady changes things. I think regardless a clear choice is to kneel on the 1 on 2nd down because it also opens up the option that the Pats will probably still allow the score on the next play to save the 40 seconds (while having wasted their last timeout), so even if you want to avoid the kick with a touchdown, you might as well do so on third down, not second.

        Reply
        1. Michael

          If you’d assume that the Giants did operate optimally in this manner, (at least waiting for 3rd down to score) then you’d have to be trading in the 15 seconds down by 1 for 50 seconds and no timeouts instead of one. What are the percentages there?

          Reply
  10. Greg Wind

    The best strategy for the Giants, given the situation of the Patriots not defending the play is to take the ball to the one foot line, protect the ball with both arms and wait for a defender to move in before crossing the line, which is functionally what happened give or take roughly a second and a half. The more you kneel, the more you can expect the defense to defend, and the larger the risk of not scoring. Eventually probability has to result in an occurrence, and if you have the choice between a single uncontested step or a multi-step process for a kick, you might be defended for taking a risk on 4 percentage points of probability for the occurrence of taking a lead that forces the Patriots to score a touchdown. The though experiment is correct, but this isn’t a situation where you get rewarded for being right 94 of 100 times. You have to be right 1 of 1 times.

    The fact that someone as occasionally clueless as Chris Collinsworth noted the strategy is all the evidence I need that strategic thinking has come a long way. The fact that Belicheck overthought this one is all the evidence I need that improving probability still requires something to occur.

    Reply
  11. Robert Meltzer

    Matt, this is a brilliant analysis! I am a diehard Giants Fan, and when Bradshaw scored, I nearly had a heart attack! Why would the Giants even risk a handoff? Eli should have knelt in the middle of the field twice (actually 3 times, there should have never been a handoff on first down). There would have been 10-15 seconds left after the field goal. I don’t want to listen to this garbage about the possibility of a bad snap. Like you said, what is the better percentage? Making what amounts to an extra point indoors, or giving Tom Brady a minute to try to beat you? I was so upset at the horrible strategy it actually took away some of the enjoyment of winning the Super Bowl!

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *