The 2017 version of the Buffalo Bills broke a playoff drought that had gone on for at least two or three years, ending the misery of long-suffering fans. They accomplished it off the back of an opportunistic defense that generated lots of turnovers. With the Buffalo Bills regressing in the win column, an often-cited contributor is the team’s lack of turnovers. Let’s explore what happened.
If you recall this piece from half a year ago, you might recall the 2017 Bills were better-than-average with takeaways—but not by a whole lot. With a league average of 22, Buffalo had 25 for the year. That was good for only ninth-best in the league. In reality, the 2017 Bills were never a turnover generator. The Bills were so “not stellar” in this category that I had this to say before the season began:
“It’s a good guess that the  Bills will remain in striking distance of this year’s  results with plenty of room to actually improve in this important metric.”
While the 2017 Bills had 25 total takeaways, the 2018 team fell off a cliff and only had 27 takeaways. Wait, that second number is bigger. You’re reading that right. The Buffalo Bills had more takeaways this season than last. They slightly improved in rank as well, going from ninth- to eighth-best.
But what if the 2018 team just had a few really awesome games that threw things off, like the Week 17 game against Miami? Surely a few four-takeaway performances could skew the data. If that were the case, while the 2018 season had more turnovers, the team was less consistent. If only we had a convenient graph that showed the number of games each regular season according to the number of takeaways.
There we go. Both seasons had twelve games in which they had at least one takeaway, as shown quickly by the “zero” columns listing four games each year. On the higher end of the spectrum, both the 2017 and 2018 Bills only had one game with four takeaways and four games each with three takeaways. There’s some difference in the one and two takeaway games. Ultimately, though, the 2018 Bills were more consistent with multiple takeaway performances (nine games to seven). Put another way, this year they were a little more “boom” but were not any more “bust.”
The team was relatively consistent with interceptions, logging 18 in Sean McDermott’s first season and 16 in the second year. The bigger swing was in fumble recoveries, with seven and eleven respectively.
Where there was a significant swing, though, was turnover differential—or the difference between takeaways and giveaways. The 2017 Bills were seventh in the league in differential. With 16 giveaways in the regular season the team was +9 in this metric last year.
Despite improving in takeaways, the Buffalo Bills fell to a tie for 23rd in the league when it comes to differential. This year’s team gave the ball away exactly twice as often for 32 giveaways and a -5 differential.
I’m assuming that last point will turn into a conversation about quarterbacks, so to help steer that debate, here’s a table to look at. The last column is the fun one. It shows how many plays should be expected between turnovers.
The first thing to note is how important quarterbacks are to turnover differential. For the two years of Bills data, quarterbacks account for 39 out of 48 giveaways. Derek Anderson was exactly as bad as Nathan Peterman. Though Peterman technically improved in his second year, this isn’t a sentence I need to finish. With Josh Allen being the clear starter going forward, his rate is the only one that really matters at this point.
Now for some context, there is a large acceptable range for good quarterback play, so don’t get too worked up about Josh Allen just yet. Picking three semi-arbitrary comparisons:
- Aaron Rodgers had only five giveaways this year for a rate of one every 202 snaps
- Russell Wilson gave up the ball nine times for a rate of 118
- Ben Roethlisberger had 18 giveaways, which makes his rate one turnover every 60 plays
Josh Allen should look to decrease how often he turns the ball over. If the Bills can put more points up, though, he might not have that far to go.