“Why are we here?” Indianapolis Colts cornerback Kenny Moore wondered back in February. He wasn’t having an existential crisis. He was trying to figure out how he and five other Colts were together at the Pro Bowl, neither recovering from a playoff run nor preparing to play in Super Bowl LVl.
It’s an annual offseason pastime for players, coaches, analysts and fans: figuring out what went wrong for their team that might, this season, go right. The simplest way? Imagine every bad break had been a good one. Just move each one-score loss into the win column and vice-versa, as Football Outsiders contributor Bryan Knowles did, and the Colts become the AFC’s No. 2 seed instead of missing the playoffs entirely.
But as I wrote in 2020, not all “one-score games” are actually close. With the advent of legal sports betting across much of the U.S., NFL fans are painfully discovering concepts like “bad beats” and “backdoor covers.” A potential nail-biter can be blown open at the very end with a turnover or return touchdown, and a couple of garbage-time touchdowns can make a blowout look close.
According to ESPN Stats & Information Group’s win probability model, 66 out of 272 regular-season games in 2021 could truly have gone either way at the finish. (As in my previous article, I defined a “coin-flip game” as having a win-probability split of 60/40 or closer at some point within the last five minutes of the game.) But just flipping the result of each coin-flip game doesn’t tell us much.
Here are the teams that would have gained the most if they’d won every “winnable” game — or lost the most had they lost every “losable” one:
|Team||Wins||Losses||Ties||Wins||Losses||Chg in W|
|Team||Wins||Losses||Ties||Wins||Losses||Chg in W|
It looks like the Colts really did get the short end of the stick: If Moore and company had won just one of the four coin-flip games they played, they would have made the playoffs. With all four switched over to the win column, they’d have finished a conference-best 13-4. The Baltimore Ravens and Minnesota Vikings were both within a few lucky bounces of 12-5 — maybe former Vikings head coach Mike Zimmer was onto something when late last season he reportedly read out a laundry list of all the times his team had gotten screwed during his tenure — and the Detroit Lions, who finished with the league’s second-worst record, could have finished 7-10. Even the Buffalo Bills went 0-4 in coin-flip games; had they split them evenly, they’d have matched the Colts’ best-case scenario of 13-4, and they’d have finished 15-2 if they had won all four.
On the other side of the ledger, the Las Vegas Raiders won five of six close games, and the Pittsburgh Steelers won four (with one tie) — making both teams’ improbable playoff qualifications seem even luckier. The NFC’s top two seeds, the Tampa Bay Buccaneers and Green Bay Packers, each won four coin-flip games. And the incongruity between the Atlanta Falcons’ win-loss record and scoring differential was obvious during the season; it’s no surprise that losing all of their close, late games would have dropped them from 7-10 to 3-14.
Baltimore is maybe the most fascinating case of all, though, since it was somehow tied at the top of both charts. The Ravens’ 6-2 start was, as I wrote in November, enabled by a string of late-game rallies that seemed unsustainably lucky. They’d go on to win just four of their league-high nine coin-flip games in 2022 and miss the playoffs completely. Playing nine total coin-flip games in a season tied the Ravens with the six-year high set by the 2018 Cleveland Browns, who broke a nearly-two-year winless streak with a Week 3 comeback over the Jets and went 7-8-1 overall, with a little help from a 4-4-1 record in coin-flip games.
Though those two squads inspired more nail-biting, hair-pulling and chain-smoking from their fans than any of the other 190 team-seasons of the last six years,1 there’s no data to suggest anything unusual is happening with the NFL and close finishes. The leaguewide share of coin-flip games has been relatively stable, between 20.3 percent and 26.6 percent of all contests, and so has the per-team average since 2016.
That said, when about one out of every four NFL games comes down to good luck in the last five minutes, the “usual” range of possible outcomes for each team is still huge. The average team’s record last season could have changed by plus-or-minus 2.1 wins depending on whether their coin-flip games went in either the best-case or worst-case direction, with eight teams having a possible swing of at least plus-or-minus four wins. Neither the Denver Broncos nor Houston Texans played any games with a 60/40 or narrower win-probability split in the last five minutes of regulation. But for every other team, there’s at least a little bit of woulda-coulda-shoulda to contemplate.
As anyone who’s played with FiveThirtyEight’s NFL predictions calculator can testify, changing individual NFL results has butterfly-effect impacts on the rest of the standings and outcomes. It’s easy to look at the best-case upside for one team and wonder “what if,” but the possibility spaces are open for all the other teams, too.
As referenced in my prior research, ESPN’s Brian Burke once calculated that NFL outcomes appear consistent with randomness driving about 24 percent of game results. That dovetails with the 23.6 percent leaguewide rate of coin-flip games over the last six seasons — though randomness doesn’t just happen in the last five minutes of an NFL game. It’s a down-to-down constant.
One could certainly call last year’s Raiders, who made the playoffs on the back of a historically unlikely series of fortunate events, lucky. But there’s another word we use to describe teams that consistently pull out wins in close situations: clutch. While Moore and the rest of the Colts’ talented 2021 squad might have been unlucky to have been watching the playoffs from their couches, they also could have used a dash of clutch performance — after all, they had four chances to win easily winnable games, and they didn’t.