Posted by Jonah Gardner on February 4, 2016
50 years ago, the NFL and the AFL Champions agreed to square off to see who was truly the best football team in the country. Today, we look back on Super Bowl I as the game where Vince Lombardi cemented his legacy, a key pitstop on football's road to becoming America's most popular sport, and a historic object of fascination. But there's one thing that we don't talk about that much: how much that game sucked.
Green Bay entered play as 14-point favorites and never trailed at any point. We have a model that uses the situation and the point spread to calculate a team's expected odds of winning at any moment in the game. Here's what the win probability graph of Super Bowl I looked like:
The Packers' chances of being history's first Super Bowl Champions dipped below 80% for just one play and roughly 20 seconds of game time. It wouldn't get much more interesting in Super Bowl II:
Here, Green Bay started with an 82% shot of winning and never went below that. Fortunately, in the course of taking 47 more shots at this thing, they've managed to get a lot more interesting. Here, for example, is what the win probability chart looked like last year:
Unless you are a Seahawks fan, you'll agree that looks a little more like what we want out of the #1 television event of the year.
Using these win probability charts, we can measure the impact every play had on the game. And, since we have one for every Super Bowl, we can go all the way back from Super Bowl I and determine what were the most important plays in the history of the game. So here they are: The 14 biggest plays in the first 49 years of Super Bowl History, as measured by win probability.
14. Super Bowl XXIII: Joe Montana pass complete to Jerry Rice for 27 yards. 32.4% win probability swing
We start with a game that was, until a decade or so ago, considered by many to be the greatest Super Bowl ever played. While the 49ers wouldn't find the end zone until 2 plays later, it was this moment, on a 2nd and 20, that would be the defining play of the drive and propel San Francisco's comeback:
Incredibly, this is the only time that either Montana or Rice will appear in the Top 14. Of course, the main reason for that is their complete dominance. The 4 other Super Bowls that featured at least one of them were (more or less) beatdowns, including this laugher the year after they beat Cincinnati:
At least the Broncos would never been blown out that badly in a Super Bowl ever again!
13. Super Bowl XXXVI: Kurt Warner pass complete to Ricky Proehl for 26 yards, touchdown. 33.9% win probability swing
The Patriots dynasty almost got held at the station because of this one:
In just about 2:00 of game time, the Patriots went from having an 86% chance of winning their first Super Bowl in franchise history to just a 25.9%, a swing of over 60 points in just 5 plays. Why so low if the Rams merely tied it? Because St. Louis was favored by 2 TDs going into the game, so our win probability model really, really liked them in OT.
12. Super Bowl V: Craig Morton pass incomplete intended for Walt Garrison is intercepted by Rick Volk. 36.8% win probability swing
The oldest play on our list comes from Super Bowl V, a game that would feature 2 players who would go on to coach in Super Bowls. It would also feature 6 INTs, including this one, which was very nearly a pick six:
Baltimore would go on to score 2 plays later and would win on field goal in the game's final few seconds. This Super Bowl would also be notable for MVP going to a player from the losing team: Chuck Howley, who had a 2-INT game. However, given that this INT would be the most important play of the Super Bowl and the most important of any Super Bowl in the 1970s, maybe they picked the wrong defender.
11. Super Bowl XXIII: Mike Cofer kicks off 58 yards, returned by Stanford Jennings for 93 yards, touchdown. 37.4% win probability swing
That's right, Jerry Rice's amazing catch in a last minute comeback drive wasn't the biggest play of Super Bowl XXIII, at least from a win probability perspective. Instead, it was Stanford Jennings, taking this kick to the house:
What's stunning about that play is how, once Jennings finds the seam in the coverage, there's basically no doubt he's scoring. As you'll see the further we go, win probability really really likes long scoring plays, a sentiment that I 100% agree with. While the Bengals had tipped over 50% at a few points in the game, this would send their win probability to 78.3%. Unfortunately for Bengals fans, all that would do is make the 49ers comeback that much more impressive.
10. Super Bowl XLIII: Ben Roethlisberger pass complete short right to Santonio Holmes for 40 yards (tackle by Aaron Francisco). 37.7% win probability swing
I'm going to be honest with you. When I went to look up this play, I assumed there was something wrong with this list. That's because my eyes went directly to Santonio Holmes' game-winning TD catch, one of the most jaw-dropping catches I've ever seen. But that play, which had a 24.9% win expectancy swing, could never happen without this one:
It's Holmes' catch and, more importantly, Francisco's slip that did more damage to the Cardinals' shot at the franchise's first championship since 1947 than the bonkers catch that Holmes would make 2 plays later.
9. Super Bowl XXXVI: Adam Vinatieri 48 yard field goal good. 39% win probability swing
Our first field goal! And it's fitting that it comes from one of the best clutch kickers of all-time
This would be the longest made FG in the 4th quarter of a Super Bowl and has the biggest win probability swing of any made FG in a Super Bowl.
8. Super Bowl XIV: Terry Bradshaw pass complete to John Stallworth for 73 yards, touchdown. 39.3% win probability swing
The next time they tell you old QBs couldn't chuck it like modern ones, show them this video
That straight-up bomb came on a 3rd and long, averted a punt on 4th, gave the Steelers a lead they would hold on to for the rest of the game, and locked up Bradshaw's 4th Super Bowl.
That's also the 2nd biggest Super Bowl win probability swing in favor of the Steelers in franchise history (we'll get to the first in a moment). That's higher than the 49ers biggest play, which we already saw at 14. Interestingly, the Cowboys, despite playing in 8 Super Bowls, do not make any appearances on this list for positive win probability plays (although they did give up the INT in #12). Neither does Green Bay. However, the most surprising omission is Denver.
The Broncos appeared in 7 Super Bowls (their 8th appearance this year will tie them with Pittsburgh, New England, and Dallas) and weren't involved in any of the Top 14 plays, on either side of the ball.
7. Super Bowl XLIII: Kurt Warner pass incomplete short middle intended for Anquan Boldin is intercepted by James Harrison at PIT-0 and returned for 100 yards, touchdown. 42.8% win probability swing
As we saw with the Holmes catch, sometimes the play that sets up the famous play is the one that ends up having the biggest impact. But sometimes, the math backs up our memory and a big play is every bit as awesome as we remember it being:
That is a 100 yard pick six coming immediately after Arizona took its first win probability lead of the game, which is probably as big of a swing as you can have on a single play, as long as you ignore how much time is left. That's why this is the only 1st Half play to make the list. In fact, the Jennings kick return was the only other play on the list besides this one that didn't happen in the 4th quarter.
By the way, how awesome was this game? So awesome that this play could happen, putting Arizona's chances of winning at just 10%, and they still mounted a comeback large enough to produce a second play on this list. In case you're curious, the Fitzgerald catch just missed the list, at 30.3%. Was this the best win probability Super Bowl ever? Hold that thought.
6. Super Bowl XVII: John Riggins for 43 yards, touchdown. 43.4% win probability swing
It's tough to make this list with a running play. In general, you're less likely to have a big play with a run, rather than a pass. If a team is losing and trying to come back, prime conditions for a win probability swing, they're probably throwing the ball. The situation has to be just right to get a run onto this list, which is why there's only one:
It also helps is you can boss a defender, like John Riggins does to Don McNeal on the way to the end zone, an MVP, and the first Super Bowl in Washington history.
5. Super Bowl XXXVIII: Jake Delhomme pass complete to Muhsin Muhammad for 85 yards, touchdown. 45.3% win probability swing
Before Jake Delhomme was a figure of ridicule, he was one of tragedy.
By passer rating, his performance in Super Bowl XXXVIII was the best by a QB in a Super Bowl loss. If you include the wins, it was 16th, higher than any Tom Brady, Ben Roethlisberger, or Peyton Manning Super Bowl. Here, he completes the longest pass on the list, down by 5, halfway through the 4th quarter to bring the Panthers back from a 6% chance of winning and give them their first Super Bowl lead in franchise history.
4. Super Bowl XXXVIII: Jake Delhomme pass complete to Ricky Proehl for 12 yards, touchdown. 49% win probability swing
And yet that wasn't even Delhomme's best play of the game. Sitting a bit outside the 10 yard line, down by 7, and with just a 19.1% chance of winning, he'd make this happen:
Delhomme is milliseconds away from what would almost be a game-ending sack and instead he finds the end zone.
Here's an interesting quirk of the win probability model. By scoring the TD, Carolina's chances of winning jumped from 19.1% to 68.1%. They would immediately drop 18 points. Why? Because they kicked the extra point to tie and force OT. The model is accounting for their odds of going for the tie and winning in OT, but it is also factoring in the generally favorable odds that NFL teams have of converting a 2-pointer.
Would it have been the right call? The Panthers had tried 2 2-pointers already and failed to convert either. But the Patriots defense was missing Rodney Harrison and had just given up 2 TDs in 8 minutes. It's a question that would ultimately not matter, since a John Kasey misfire on the kickoff, Tom Brady's brilliance, and another clutch Adam Vinatieri kick would seal the game for New England.
Was this the best Super Bowl ever? It doesn't even crack this NFL Network Top 10 and I think that's the case for most people, depending on your age and level of rooting interest in this game. But consider it had 2 plays that made the top 5 and a 4th quarter win probability chart that you have to be at least 54" tall to look at:
Also, let's take a moment to appreciate Ricky Proehl, who, according to win probability, is the most clutch WR in Super Bowl history. He joins Kurt Warner and Jake Delhomme as the only players to show up on this list twice.
3. Super Bowl XLIX: Russell Wilson pass incomplete short right intended for Ricardo Lockette is intercepted by Malcolm Butler at NWE--1 and returned for 3 yards. 58.8% win probability swing
It isn't just recency bias. The last non-QB kneel play to take place in a Super Bowl, as of this writing, was the 3rd biggest of all time.
Let's start with this: the Patriots had a 93% chance of winning just 1:46 of game time before this happened. However, after quickly moving the ball from their own 20 to the Red Zone (a drive fueled mainly by a Jermaine Kearse 33-yard catch worth a 21.1% shift in win expectancy) the Seahawks were in striking distance.
After Lynch's 4 yard run to set up 2nd and Goal at the 1, New England's chances of victory had dwindled to 41.2%, meaning that, depending on how much time was left when they scored, Seattle was on the verge of a play that would crack the Top 10 of this list on its own right. Instead, we got the biggest defensive play in Super Bowl history, a walkoff INT that turned sub-coin flip odds into a 5th New England title.
By the way, I checked and there've been 82 INTs on X & Goal from the 1 or 2 since 1994. Of those, 28 came in the 4th quarter, 4 were in the playoffs, and 13 came when the team with the ball trailed by 4-8 points. Only one play since 1994 satisfies all four of those criteria and its this one.
2. Super Bowl XXV: Scott Norwood 47 yard field goal no good. 61% win probability swing
Bills fans, don't do this to yourselves. Don't watch this clip. Don't read this section. Just go ahead and skip to #1. Resume whatever you were doing with your day before you had think about this play:
I have an aesthetic qualm, completely removed from the numbers, with putting this play this high. On the other ones, someone is doing making the play happen. Malcolm Butler picks off a pass or John Riggins breaks a monster run. They actively change the course of the game in their team's favor. Here, no one does anything to cause the miss; poor Scott Norwood just misses. He misses a kick that win probability thinks he has a roughly 60% chance of making.
Also, do you want a little "Lincoln had a secretary named Kennedy" type weirdness? There are two field goals on this list of the 14 most important plays in Super Bowl history. In one, Scott Norwood misses a 47-yard FG as time more or less expires, starting off a series of historic championship disappointment for the Bills. In the other, Adam Vinatieri makes a 47-yard FG as time expires, starting a dynastic run of success for Buffalo's divisional rivals, the New England Patriots, and their coach Bill Belichick.
The defensive coordinator for the Giants who came up with the game plan that slowed down the Bills' up-tempo attack, kept the game close, and set up this moment? Bill Belichick.
Also, did you notice the play-by-play announcer on both this play and the one above it? Or the Cardinals-Steelers game, for that matter? If there's a historic sports moment, Al Michaels is going to be there.
1. Super Bowl XLII: Eli Manning pass complete short left to Plaxico Burress for 13 yards, touchdown. 61.9% win probability swing
Our win probability model doesn't care about your helmet catches, sack evasions, or (if you're a Patriots fan) uncalled holding penalties. The Tyree catch represented a 17.5% uptick in the Giants' win probability; Plax's TD catch more than tripled that:
The magic of the Tyree catch shows up in our play-by-play as "Eli Manning pass complete deep middle to David Tyree for 32 yards (tackle by Rodney Harrison)." Even after making that play, the Giants still had to move the ball 24 yards against one of the best teams in history in under a minute, something that the win probability model gave a 1 in 3 shot of happening.
Keeping in mind that the Patriots were favored by 12.5, giving them an 81.4% chance of winning at kickoff, that the Giants had just a 5.3% chance of winning when New England kicked off to them to start that drive, and that this is the capper on arguably the most famous TD drive of the last 30 years, this seems like a good choice to me. It's also interesting that, while some of these choices have been surprising, the three plays everyone would pick for the Top 3 were the Top 3, and the only ones to represent a swing of over 50% in win probability.
So, if it feels like Super Bowls have gotten more entertaining in recent years, the numbers kind of back this up. 8 of the top 14 plays took place after 2000 and 4 were within the last 10 years. The most recent Super Bowl play was the 3rd biggest of all time!
Given how many of these plays are passes (11 out of 15, counting the 2 INTs), maybe football's shift to being a passing-dominated game has also made it a less stable one, with more of a tendency to be decided by late drives and exciting plays. And maybe we'll be lucky enough to see another addition to this list on Sunday.
Was Manning to Burress your favorite Super Bowl play? If not, what was? If a play didn't make the list that you're curious about, you can look up the win probability change in that game's box score.