≡ Menu

Interceptions By Win Probability, Part 2

Mayfield is having a tough start to the year, but don’t point to his interception against Tennessee as a reason why.

Last year, I wrote about how not all interceptions are created equally, and that there is a large diversity in the impact of interceptions.

There have been a lot of meaningless interceptions this season already. Baker Mayfield threw an interception down by 23 points with 3 minutes remaining at the Cleveland 20. That interception dropped the Browns chances of winning from zero to zero. Trailing by 42 points, Josh Rosen opened the 4th quarter of the opening game loss to the Ravens by throwing a meaningless interception. The next week, Miami’s Ryan Fitzpatrick threw two 4th quarter interceptions in a blowout loss to the Patriots: yeah, maybe all Dolphins interceptions are meaningless, but these three were *really* meaningless. And the most meaningless (is that a thing?) one yet? Rosen threw an interception on the final play of that Patriots game, too.

Using the data from scrapr, I calculated the win probability cost of every interception this year. [1]Note that there are some bugs in the data. A handful of the INTs have the win probability backwards, or are just calculated clearly incorrectly. I manually checked these to revise. Also, for the … Continue reading

The most meaningful interception of the year came when Aaron Rodgers threw an interception (which probably wasn’t his fault) on Thursday night. Green Bay trailed by 7, with 2nd-and-3 at the Eagles 3-yard line with 28 seconds remaining. How likely was it that Green Bay would win the game at that point? ESPN put Green Bay’s win probability at 38%, which feels about right to me, right before that play. The interception, of course, ended the game, making it a 38% decline in win probability.

The other two most meaningful interceptions of the year also involved NFC North defenses. The Packers gained 26% of win probability on an interception of Kirk Cousins in week two. Minnesota trailed 21-16 with 5:17 remaining, and had the ball on 1st down at the Packers 8-yard line. The game was a toss-up at this point, but after the interception, the Packers had 1st and down, up 5, at their own 20-yard line with five minutes left. That boosted Green Bay’s win probability from about 51% to about 77%.

The other one came from Philip Rivers, who threw an interception to seal the Chargers’ fate against the Lions. Los Angeles was down by 3 with 1:10 remaining and at the Lions 28-yard line. It was 3rd-and-19, so the Chargers were probably hoping for a field goal, a defensive stop, and then to win in overtime. Los Angeles still had a small chance to win after this (they had all three timeouts and were only down 3 points with 63 seconds remaning), but this reduced their chances of winning from 41% to 12%.

The graph below shows all 71 interceptions so far this season, and how much win probability was lost on each one.

Jameis Winston has cost his team the most win probability on interceptions this year: he has thrown 4 interceptions but they have totaled 54% of negative win probability added. The table below shows the amount of win probability lost on all interceptions this year, and on the average interception:

Once again, the win probability data comes from Ron Yurko and the NFL’s scrapr data. There will be a few bugs, and we can disagree with some of the determinations (more on this in the comments). But this is a pretty good start to remember that not all interceptions are created equally.

References

References
1 Note that there are some bugs in the data. A handful of the INTs have the win probability backwards, or are just calculated clearly incorrectly. I manually checked these to revise. Also, for the Rodgers interception, I looked at the ESPN Gamecast to derive the win probability.
{ 0 comments }