Curran: Patriots Super Bowl comeback was more probable than not

Was what the Patriots accomplished Sunday unlikely? Yes.

Was the likelihood of it happening as infinitesimally small as is being touted by stat wonks and columnists? No.

Given the number of times the Patriots have taken a flamethrower to win-probability charts since 2012, it would have been more unlikely had the Patriots not made a run that brought them within striking distance at the end.

Win-probability doesn’t take into account the abilities and histories of specific teams and individuals. When Team X trails by Y points with Z time remaining … I don’t know how they figure the stupid thing. But I do know that who X is matters.

If you bind the hands and feet of 100 people and throw them in a pool, the survival rate will be low. If the one that survives happens to be a Navy SEAL, well, we’d say that guy was an exception. He was trained for that. His making it out was totally understandable.

The analogy extends to the Patriots. Starting with their 16-year veteran starting quarterback playing in his seventh Super Bowl, their head coach with 42 years NFL experience and coaching in his 10thSuper Bowl, they were trained for this in a way that nobody – not Walsh, Montana, Unitas, Noll, Bradshaw, Shula, Manning or Otto Graham – ever were.

They’ve done this stuff before.

Consider this. If in the last Super Bowl, the Patriots win probability in the fourth quarter was as low as 3.5 percent and in this one their probability was .4 percent and they won both, does a standard probability measure apply? Less than a five percent chance of winning both times. 100 percent success rate.

And those games weren’t outliers. Go through recent seasons. The number of times they’ve either snatched victory from the jaws of certain defeat or come stupidly close to doing so is evidence that what happened in Super Bowl LI – as momentous, stupendous and incredible as it was – wasn’t a lightning strike event. For them.

In 2012, the Patriots trailed the San Francisco 49ers 31-3 with 10:21 remaining in the third quarter. The folks at the incredible football resource site, Pro Football Reference, didn’t start charting win probability until 2013, but using their calculator, the Patriots win probability was .10 percent. New England tied the game at 31 with 6:45 left in the game. They fell behind 41-31 after the Niners kicked a field goal with 2:30 remaining following the Patriots failed fourth-down attempt at their own 12. The Patriots still got the ball back, kicked a field goal to account for the 41-34 final and were an onsides kick recovery away from having a chance to tie it.

The 2013 season was loaded with win-probability refutations.

In Week 6 against the Saints, the Patriots trailed 27-23 with 2:29 after a New Orleans field goal. Brady was picked on the first play of the ensuing drive. Saints ball. Win probability 8.27 for the Patriots. New England won with a Brady to Kenbrell Thompkins touchdown pass with 10 seconds left on a drive that included a fourth-and-4 conversion to Austin Collie. Articles were written about how statistically improbable it was. Later that night, Big Papi did this. Also, improbable.  

In Week 11 at Carolina, the Panthers’ win-probability was at 99.60 when Cam Newton put his team up 24-20 with a 25-yard touchdown pass to Ted Ginn. And it was still at 93 percent after Brady missed Shane Vereen to bring the Patriots to a fourth-and-10 with 51 seconds left. Then Brady hit Gronk for 23 and a conversion. Carolina’s probability sunk to 62.49 with six seconds left when the Patriots had a last shot at the end zone. Brady was intercepted but the obvious defensive pass interference in the end zone that went uncalled would have put the Patriots at the Panthers 1 for a final play to win. 

The following week against the Broncos, the Patriots were down 24-0 at halftime. Win probability? .20. They came back to win 34-31 in overtime.

Week 14 against the Browns was probably the most absurd comeback. Trailing 26-14 with 2:43 left after a Browns touchdown pass by Jason Campbell, win probability was .20. Then Brady threw a touchdown to Julian Edelman with 1:04 left, the Patriots recovered an onside kick, there was a sketchy pass interference called on Cleveland on a pass intended for Josh Boyce and Brady threw a touchdown pass to Danny Amendola with 35 seconds left.

In 2014, two of the three Patriots postseason wins defied probability. The Ravens chances of winning were 90.20 when they went ahead 28-14 in the third quarter. It was still at 73 percent in the third when the game was tied at 28 and the Ravens had first-and-10 at the Patriots 25 with 11 minutes left. The Patriots won 35-31 on a Brady to Brandon LaFell touchdown pass with five minutes left.

In the Super Bowl, Seattle was 96.5 percent assured of winning when they went ahead 24-14 in the fourth. Interestingly, Seattle – which is also a bit of an outlier from normal with brilliant quarterback and experienced coach – fought back from having only a seven-percent chance of winning when the Patriots went up 28-24. After Jermaine Kearse’s juggling catch, the win likelihood for the Seahawks was 63 percent. Didn’t work out for Seattle.

In 2015, the Patriots’ win probability was .20 when it was 35-14 with 5:25 left in the game against the Eagles. The Patriots scored twice to make it 35-28 with 3:02 left before sputtering.

In the 2015 AFC Championship Game – for the third time in four postseason games – the Patriots had a win-probability of less than one percent when they failed on a fourth-and-6 at the Denver 14 with 2:25 left while trailing 20-12. Yet they got a stop and got the ball back. They still had a less than four-percent chance of winning when facing fourth-and-10 with the 50 with 1:39 left. And Brady hit Gronk for 40. After their touchdown made it 20-18 with 12 seconds left, they had a 39.03 chance of winning. They died when Brady’s two-point conversion pass was picked.

This season, they only had one occasion before Sunday where they were dead and buried and almost kicked the lid off the coffin. Against Seattle, the Patriots had just an 11 percent chance of winning after Edelman was stripped and the Seahawks turned the turnover into a touchdown to go up 31-24 with 4:30 left. And New England still got it to the Seattle 1 with four shots to tie.

Super Bowl LI stands out because no game was bleaker for longer. In that way it stands out. But having just listed 10 examples in the past five seasons (for a team that rarely trails), what other evidence is necessary?

Anyone that didn’t expect Falcons to have to sweat just a little in the final 15 minutes hasn't been paying attention.

You can believe that the men who built the Atlanta roster – Thomas Dimitroff and Scott Pioli, former Patriots executives  – were as unsettled as anyone. They knew the avalanche could be started by the most delicately descending snowflake. Even if the chart didn’t.

So the needle stayed at 99.80 probability for the Falcons with 1:30 left in the third quarter, but the Jake Matthews holding penalty that moved the Falcons out of field goal range was that flake. An incompletion on third down, a punt, a field goal drive and still the Patriots had just a 1.20 percent chance of winning. According to the chart. And it was still at 1.20 after Dont'a Hightower’s strip sack at the Atlanta 25 with the score 20-12.

We’ve all watched, read and digested ple nty of what unfolded from there. For the uninitiated, it was unimaginable.

For a region of fans lucky enough to follow this team at this time – and the even more fortunate media that’s had a front-row seat and access to a historic run – it was just what they do.

Win probability is a fun tool, a guide, a measuring stick. But it’s not football analysis. It’s number crunching.

Did I think it was likely the Patriots would win when it was 21-0 in the second or 21-3 at the break? Can’t say I did. But I knew it would get close.