In week 1 last year, Tampa Bay quarterback Ryan Fitzpatrick had a game for the ages. He averaged 17.75 ANY/A on 28 dropbacks, and was the catalyst in the Bucs 48-40 upset over the Saints. Tampa Bay’s running game underperformed, and its defense was dreadful, but Fitzpatrick and the passing offense were so effective that the Bucs won anyway.
Here is the Expected Points Summary from that game, courtesy of PFR.
Passing is king in football, and you won’t be surprised to learn that in lots of games, the passing game is the reason a team wins. Even in games where a player throws a pick six — say, this Chargers/49ers game from September — strong passing the rest of the way can make up the difference. In this game, Los Angeles won by 1 point and finished with +2.2 points of passing EPA, as Philip Rivers still finished with 6.43 ANY/A.
In total, there were 113 games last year where a team won and their passing EPA was larger than the margin of victory. This is in stark contrast to yesterday, where there 23 games where the team that won had a rushing EPA larger than the margin of victory. This means that in 44% of games, the team that won would “have lost” if they had an average passing attack.
That’s pretty significant, coming in nearly 5 times as large as the rushing data. That’s obviously an important finding, but it’s also probably not too surprising to most people.
What is surprising is what happens when we look at things from the other perspective. There are lots of games with bad passing metrics, but that doesn’t mean that an average passing performance would have flipped the result. You may not remember that Derek Anderson started a game for the Bills and was awful, but Buffalo’s passing game produced -24 points of EPA in a game they lost 37-5.
On the other hand, you are probably thinking of games like this Blake Bortles disaster against Tennessee last September. Jacksonville’s passing attack finished with -7 points of EPA, and lost 9-6 to the Titans. So with an average passing game, the Jaguars “should” win that game.
The surprising thing is that only 35 games were impacted this way, or 14%. That’s significantly lower and hard to explain away. One issue, however, is that PFR’s EPA data is not era-adjusted. As a result, 67% of the time last season, a team finished with a positive EPA number. The average EPA from a passing game was 4.64 points. And even if EPA data was era-adjusted, there is still the thorny issue that passing plays have more positive EPA than rushing plays, so you might always expect an average passing game to have positive EPA (unless you treat passing differently than rushing, which I don’t think is advisable).
Even still, that doesn’t explain the large disparity between above-average passing being responsible for winning (44%) and below-average passing being responsible for losing (14%). But if we split the difference, we can say that on average, 29% of the time, a team’s passing offense flipped the outcome. And yesterday, we saw that 10% of the time, a team’s rushing offense flipped the outcome. This would imply that pass offense is about three times as important as rushing offense. This study does not prove that, of course, but that result is consistent with what I would expect to see if we ran a study to prove that.