≡ Menu

Last year, I wrote a post on the plays that had the biggest impact on the eventual Super Bowl champion. These were the plays that affected the Super Bowl win probability by the biggest amount among teams that did not win the title. At the time, the Buffalo Bills were on the short end of the most influential play in the Super Bowl era. When Frank Reich put the ball down for Scott Norwood, I estimated that the Bills had a 45% chance on winning the Super Bowl.1 After the kick went wide right, the Bills’ win probability fell to zero. The 45 percentage point fall was the biggest change for a non-champion of any play in the Super Bowl era. Over 48 years, a bunch of plays fell in that range, but no team could point to a single play as having lowered its championship chances by so large an amount.

A couple weeks ago, that long-held record got broken kind of like Michael Johnson broke the 200-meter record in the Atlanta Olympics. Malcolm Butler’s pick obliterated the old mark. My estimate has the Butler interception as increasing the Patriots’ chances of winning by 0.87. There is no doubt that what some have called the Immaculate Interception is on an island by itself as the most influential play in NFL history.

To get that change in win probability from Butler’s play, I am going to assume that the Seahawks would have run on third and fourth down. I am going to give a run from the one a 60% chance of working. That might seem high, but the Patriots were the worst team in football in stuffing the run in important short-yardage situations either on third or fourth down, or down by the goal line. And their limited success mostly came against terrible running teams. It is not a huge sample, but against teams outside the worst quarter of rushing teams by DVOA, the Patriots had allowed opponents to convert 16 of 17 times with two yards or less to go for a first down or touchdown. If we add the playoffs, they actually had three more stops against good running teams (Baltimore and Seattle), albeit in games where the opponent had a good amount of success on the ground.2 With Seattle being the best rushing team in football by a mile and the Patriots being at best not great in run defense in that situation, it seems hard to think that Seattle had anything less than a 0.60 chance of scoring on a run.

Here are my other assumptions:

  • Give Seattle a 2% chance of throwing an interception on the slant.3
  • Give Seattle a 0.5% chance of fumbling on a running play.
  • The slant had a 55% chance of scoring a touchdown.
  • The Patriots have a 5% chance of winning if the second down play scored the touchdown with a snap at 0:26, 3% if it was second down, and 1% if fourth down.4

Under those assumptions, the Seahawks’ win probability at the snap―taking the slant call as given―was 0.87.5

Prob (Score 2nd down)*Prob (Win if Score) + Prob (3rd down happens)*Prob (Score 3rd down)*Prob (Win if Score) + Prob (4th down happens)*Prob (Score 4th down)*Prob (Win if Score) =

0.55*0.95 + 0.43*0.60*0.97 + 0.17*0.60*0.99 = 0.874

One interesting thing to note is that Pete Carroll’s seemingly odd explanation about wasting a play is not quite that crazy. If the Seahawks had just thrown the ball away on second down to waste some time, but leave them the timeout to run twice, their win probability was 0.82. That fall is small in part because the Patriots’ chances of winning after a Seattle score get even lower if they score on a later down. It’s still not a great idea because Marshawn Lynch’s chances of scoring on third or fourth down are not 100%, but Carroll may have been thinking along these lines.6

Two other plays from this postseason deserve special note. First, the Jermaine Kearse Manna-from-Heaven catch could have been a top five most-influential play in NFL history. Under my rules, it does not get on the list because a game can only have one play on the list and the team that it hurts has to not win the Super Bowl in the end. But if Marshawn Lynch ran it in and the Seahawks won, the Kearse catch would have been the play that hurt the Patriots more than any other. Brian Burke’s Win Probability Calculator assigns the Patriots a 62% chance of winning at the snap and a 25% chance of winning after the completion. So the calculator estimates that the Kearse catch lowered the Patriots’ chances by 37%, eerily similar to the 39% hit the Patriots took from the Helmet Catch.

While that play does not make the cut because the Immaculate Interception wiped it out, another play from this postseason almost did squeeze onto the expanded list. The onside kick recovery by Seattle in the NFC Championship game, according to my estimates, lowered the Packers’ chances of winning the Super Bowl by about 14 percentage points. Here are the ingredients for that calculation:

  • Give Green Bay a 0.8 chance of recovering the onside kick
  • Green Bay has a 0.95 chance of winning if they recover
  • Green Bay has a 0.55 chance of winning if they don’t recover7
  • Green Bay would have had 0.45 chance of beating the Patriots in the Super Bowl.

So, before the onside kick, the Packers’ chances of winning the Super Bowl were:

[Prob(Recover onside kick)*Prob(Beat Seattle if recover) + Prob(Don’t Recover)*Prob(Beat Seattle if don’t)]*Prob(Beat New England) = (0.8*0.95 + 0.2*0.55) * 0.45 = 0.39

After the onside kick, the Packers’ chances of winning the Super Bowl were:

Prob(Beat Seattle after not recovering) * Prob (Beat New England) = 0.55*0.45 = 0.25

So the failure to recover the onside kick lowered the Packers’ chances of winning the Super Bowl by 14 percentage points, leaving the Brandon Bostick botched recovery just off the list.

Below is the updated and expanded-to-25-plays list of the most influential plays in NFL history. The full set of ground rules can be found in last year’s post. The SBD, or Super Bowl Delta, value refers to the change in the probability of winning the Super Bowl for the team on the short end of the play. Since Norwood’s miss lowered the Bills’ chances of winning the Super Bowl by 45 percentage points, I assign an SBD value of 45 to that play.

RkNFL YearGamePlayTime at snapLosing teamGame WP ChangeSBD value
12014Super BowlImmaculate Interception00:26Seahawks0.8787
21990Super BowlWide Right00:08Bills0.4545
32008Super BowlHolmes 40 yards to Arizona 601:02Cardinals0.4242
42007Super BowlHelmet Catch00:45Patriots0.3939
51982Super BowlHere Comes the Diesel10:28Dolphins0.3636
62012Super Bowl4th down incomplete to Crabtree01:5049ers0.3535
71988Super Bowl27 yards to Rice01:15Bengals0.3434
81979Super BowlFirst pass to Stallworth12:15Rams0.3131
92013NFC ChampionshipSherman tip-to-INT vs. Crabtree00:3049ers0.528
101970Super BowlMike Curtis INT of Morton01:09Cowboys0.2727
111967NFL ChampionshipStarr QB sneak00:16Cowboys0.3525
121993Super BowlJames Washington fumble return14:34 in 3rdBills0.2424
131972Divisional RoundImmaculate Reception00:22Raiders0.9524
142011Super BowlWes Welker drop04:00Patriots0.2323
152011AFC ChampionshipLee Evans drop00:28Ravens0.4523
161990NFC ChampionshipRoger Craig fumble2:4049ers0.4523
171987AFC ChampionshipThe Fumble01:12Browns0.422
181999Super BowlWarner to Bruce for 73-yard TD02:05Titans0.2121
191981NFC ChampionshipDanny White fumble after The Catch00:38Cowboys0.4221
201975Super BowlWagner INT of Staubach08:41Cowboys0.2121
211975Divisional RoundHail Mary00:32Vikings0.8520
222001Super BowlBrady to Brown for 23 yards00:29Rams0.220
231978Super BowlThe Sickest Man in America2:46 in 3rdCowboys0.1717
242010Super BowlMatthews forces Mendenhall fumble15:00Steelers0.1717
252012AFC DivisionalFlacco-to-Jones over a sleeping Rahim Moore02:13Broncos0.1717

On this updated list, Butler’s pick looks like Tiger Woods’s -12 on top of the scoreboard at the 2000 U.S. Open. The Immaculate Interception stands alone, almost twice as influential as any other play in the Super Bowl era.

  1. Recent research by Chase suggests something similar. []
  2. Note that the stop against Baltimore should not even count. In an otherwise great game for Gary Kubiak, he called for a reverse to Michael Campanaro on third-and-1 in the second quarter. The run was stopped for a loss. The Patriots basically could not stop Justin Forsett, making the reverse call very unnecessary. []
  3. You could argue this is too high. It’s not going to make a big difference, but I do think the chance of the pick on that route is higher with Russell Wilson height making it even a little more likely that the throw gets deflected on the line. I would not have run the fade either, with its low success rate. I would have faked a handoff to Lynch and rolled Wilson right with a run-pass option. Of course, I would have run three times, snapping a little earlier if need be. []
  4. Really think these might be a little high, if anything. []
  5. There’s just one thing to add here. You would want to account for the possibility of a run losing yardage. To keep things not too complicated, I have not done that here. But my other assumptions are charitable to the Patriots’ chances, so hopefully this just balances out. []
  6. Chase note: It’s also crazy because the “not wasting a play” idea was self-inflicted.  Seattle chose to snap the ball with 26 seconds left, which makes me think that Carroll is just coming up with a narrative, rather than being forthright.  If Seattle wanted to run three times, they could have snapped the ball with 30 or 35 or 40 seconds remaining; knowing that the team wanted to pass on 2nd down, it made sense to milk the clock down inside of 30 seconds. []
  7. Burke’s calculator estimates this as 0.64. That seems too low to me, so I am adjusting that number a bit. []
  • Andrew Healy

    Responding to Chase’s note. The wasting a play idea means you want to run the clock down as much as possible. Then you run it in on third or fourth down to give the Patriots almost no time to respond. So just throwing out the idea that Carroll might have been thinking about limiting the Pats’ time even lower than 0:21 or whatever would have been left after second down. I don’t think this adds up, but I think there’s maybe some logic there. Of course, I am just guessing and may be way off.

    • I agree they could have run 3 plays even with 26 seconds left. But I think people have gotten their analysis backwards. It wasn’t:

      There were 26 seconds left —> Therefore the Seahawks must pass once out of 3 times

      It was

      Seattle chose to pass on second down —> Therefore the Seahawks drained the clock to 26 seconds

      If the Seahawks wanted to run three straight times, snapping the ball with 30 seconds left would probably have been optimal. Lynch gets tackled on 2nd down with say, 25 seconds left, call timeout with 23 seconds left. Then you basically know you are going to call two running plays, and use the timeout to get everyone on the same page. Run on 3rd down, and if he gets stuffed, you have 18 seconds to line up everyone and run on 4th down. Not too hard.

      • Andrew Healy

        That’s a neat point. And I’m right with you on the right time to snap to get three runs:

        https://twitter.com/AndHealy/status/562331002836512768

        I certainly like three runs there. Although one of those probably would be a Wilson run with a pass option.

  • Bob Harris

    Just out of interest, where would The Tackle rank on this list? Obviously it must have been lower than the Warner to Bruce TD, but would it have made the list otherwise? I guess the fact it would only have taken the game to OT would count against it.

    • Well, after the tackle, the Titans WP dropped to 0. Had they scored a touchdown, Tennessee’s win probability would have been close to 50% (I suppose the possibility of a missed XP + being the slightly weaker team makes it just south of 50/50). So I think you could argue that given how the play started, the tackle had an SBD of about 50.

      But that’s not how Andrew is measuring SBD: he’s measuring it from the start of each play, not the middle. So the question would be, what was Tennessee’s chances of scoring a touchdown on that play? I would guess around 25%, which would imply a WP of about 12%, and an SBD of 12.

      • Bob Harris

        Cheers, I haven’t quite got to grips with the win probability calculator so couldn’t work out where the Titans stood before the play. As you say, the options afterwards were either 0 or 50 (more or less), so it all comes down to the likelihood of them getting the touchdown on that play.

        • Andrew Healy

          25% is about exactly what I was thinking. The calculator spits out 37% whether it’s the five-yard line or the ten at the beginning of the play. One of the cases where the calculator just doesn’t work very well.

          • Bob Harris

            Cool, I got 37% as well, but thought that seemed very high. Also the fact that the win probability changed considerably between it being 1st, 2nd or 3rd down seemed a bit odd with only six seconds remaining.

  • Tom

    Andrew –

    This is awesome. I was keeping a list like this myself, using Brian’s WP calculator and just thinking of plays from memory, etc.; great to see something definitive!

    I believe you do need to make one correction – you have the Helmet Catch at #4, but I think that should be the Steve Smith catch for 12 yards which happened two plays later (WP calculator has that play as +0.38 WPA). The time at the Helmet Catch snap was 1:15, the time at the Smith Catch snap was 0:45. Per the WP calculator, the Helmet Catch had a WPA of +0.24.

    I wouldn’t have “caught” this, except that when I was fiddling around with this stuff, I found it interesting that that Smith catch was actually “bigger” than the infinitely more exciting, and iconic, Helmet Catch.

    Forgive me if you’ve somehow figured in the incredible circumstances of the Helmet Catch play into the calculation!

    Again, awesome post, thanks Andrew!

    • Richie

      Recently I was surprised to discover that the Helmet Catch was a third down play. For some reason my memory was that the play was on fourth down. When Eli almost got sacked before throwing the ball, I remembered thinking that a sack would end the game. Damn memory.

    • Tom

      Gotta reply to my own post. Andrew, just read your previous “One Play Away” post, and I see that you do figure in the actual circumstances of each play, as opposed to a simple “snap to snap” WP calculation. So I retract my statement regarding the Helmet Catch not being number 4! Apologies to all for muddying the waters!

    • Andrew Healy

      Thanks, Tom. Appreciate the kind words. Was just about to bring up exactly what you saw in the old post!

  • Tom

    One more thing – your descriptions of the plays are freaking great. “The Sickest Man in America”, “Here Comes the Diesel”, and my favorite – “Flacco-to-Jones over a sleeping Rahim Moore”. Good stuff.

  • Rick

    Seems weird that 7/25 (i.e. 28%) have occurred in the past 5 years. But then again, for most of the 80s and 90s, the Super Bowls were blowouts.

    • Andrew Healy

      Yes, I almost wrote about that. We are in a very good era for important plays. Mostly driven by close Super Bowls.

    • Tom

      Rick –

      That’s one of the cool things about a list like this – it’s highlighting the fact that we’ve had some great Super Bowls in the past 10 or so years. I remember those train wrecks of the 80’s and 90’s….SF 55, DEN 10….(shudder). It still fascinates me that if there was one game that most of us thought would be great, it was SEA-DEN, and man was it bad. Guess that’s why they play the games…

    • Tom

      Rick –

      Some numbers regarding those Dark Years:

      From 1984 to 1993, the average margin of victory in the Super Bowl was 23.6 points. And that’s including two games that were decided by 4 (1988, SFO 20, CIN 16) and 1 (1990, NYG 20, BUF 19).

      In comparison, the last 11 games have an average MOV of 8.9, and that’s including last year’s 35-point blowout.

      Considering the above, and the fact that we just witnessed the single most impactful play in the history of the game, I’d say we’ve been spoiled.

  • Independent George

    I know it’s outside the scope of your rules, but what about the dropped INT against Joe Montana in the 4th quarter of Super Bowl XXIII? I can’t even remember the name of the DB that did it.

    The other play I’m wondering about was Earnest Byner’s fumble against… crap, I can’t even remember. Damn you memory!

    • Andrew Healy

      Yes, the Billups play loses out to the 27-yard pass to Rice for that game, which happened later in the game. Part of the reason that it loses out is because the Bengals were already ahead at the time, and so catching the pick could only increase their WP so much.

      The Byner fumble was against the Broncos. That one is really close to making the list. The problem is that the Browns only had something like a .3 chance of winning the game at the snap and then they need to win Super Bowl after that.

      • Richie

        I think you guys are mistaken about Lewis Billups. He didn’t drop the interception. Joe Montana willed the ball to not be intercepted. Because Joe Montana is a winner, and does not make mistakes in big games. He knows how to get it done in pressure situations.

        • People forget that.

  • David Condon

    What about the Bryant drop or the Flacco interception from the divisional games this year? How do those stack up?

    • Daniel Saunders

      That was a catch. And I’m not a Dallas fan…at all.

    • Andrew Healy

      Just to ballpark those: Flacco pick probably lowered their chances from 35% or so to zero. Give them a 50% chance of winning the next two games, and SBD is about 9. Bryant drop would have swung their WP by maybe 30 pts. They would have been underdogs in the next two games, so SBD would be lower. A divisional round play has to be a very big swing (like the Immaculate Reception) to have a shot at making the list.

  • Patrick Allison

    Looking at the recent games here, going backwards, there are interesting things to note in the games that aren’t here. I’m using PFR’s win probability, which is not terribly fair here, since it adjusts things based on Vegas lines. Burke’s win probability calculator isn’t working right now, sadly.

    If we look at plays in the 4th quarter that rate, by PFR, higher than ~17% or so:

    Missing Super Bowls:
    Year, score differential:
    2013: 35 (duh)
    2009: 14 (Matt Stover’s missed 51 yard FG rates pretty high by PFR)
    2006: 12 (Grossman’s pick-6 at the beginning of the 4th quarter rates pretty high)
    2005: 11 (the infamous Hasselbeck Low Block rates pretty high)
    2004: 3 (*nothing*)
    2003: 3 (John Kasay’s kickoff out of bounds was *huge* – I’m surprised it doesn’t make this list. It’s a boring play, but hey, it got rated the 6th Biggest Fail in Super Bowl history. PFR has it as a 28% swing.)
    2002: 27 (duh)

    2004 – SB 39, Eagles-Patriots – is interesting to me because the game was tied through 4 quarters, and then became a gimme for New England not 5 minutes into the 4th Quarter. But *none* of those plays were significant game-changers. None. They were all low-percentage point successes. The biggest play was probably a 19-yard pass with a 15-yard personal foul, but even that was only like, 8%. It’s like it was the most boring close Super Bowl ever.

    • Andrew Healy

      I know that I evaluated the Kasay kick out of bounds. I think I probably gave the Pats something like a 65% chance of winning before the kick and about a 75-80% chance afterwards. Note that the PFR WP numbers are often good, but occasionally pretty screwy. My favorite case of this was the 99% chance BUF had before Norwood’s kick:

      http://www.pro-football-reference.com/boxscores/199101270buf.htm

      • Tom

        Yes, the PFR numbers are a little wacky. Somewhere on the site, Doug (I think) goes over how they came up with their model. It’s based on an Excel formula from Wayne Winston’s “Mathletics” book, plus Expected Points. Actually, it works fairly well, especially in the first 3 quarters…I’ve checked it against Brian’s calculator a couple of times, and they’re relatively close. But yeah, in the 4th quarter, and specific field goal and goal line situations, it falls apart. One of the bigger issues, among other things, is that the formula thinks there’s a difference between being 1 point down and 2 points down near the end of the game when a field goal will win it either way.

  • Tom C

    Whee is the Music City Miracle?

    • Andrew Healy

      Can’t really make the list from the wildcard round because they would have had to win three more games. And the Bills would have been underdogs in all three, so the SBD is under ten. If they had a .5 chance of winning each game, then SBD would have been just under 0.5^3 = 0.125. Just under because change for the game from the MCM was < 100%.

  • Scott Kacsmar

    I’ve never seen the video of this play, but this is an interesting one likely only remembered by fans of the two teams.

    1987 NFC Championship Game – Vikings at Redskins
    Down 17-10, Wade Wilson’s pass on 4th-and-4 from the WAS 6 was dropped by Darrin Nelson at the 1-yard line with 52 seconds left.

    That WAS team lived a bit dangerously in the postseason. They trailed 14-0 in Chicago before a 21-17 comeback win, and of course erased the 10-pt deficit in the SB with that massive 35-point quarter.

    • Andrew Healy

      That WAS team wasn’t the 1991 team, either. They were dogs to Denver in the Super Bowl. Note that the SBD for the Nelson pass was basically a little less than half of The Tackle. Similar situations, except WAS would have had time to score before overtime and MIN would have had to win another game if they’d beaten WAS.