Against the Jets in week 5, trailing 34-16 with 11 seconds left, Case Keenum threw an interception on the game’s final play.
Against the Broncos in week 1, the Seahawks had the ball with 2 seconds left at their own 11-yard line, down by 3. Russell Wilson threw an interception on the game’s final play.
Against the Panthers in week 3, the Bengals were down 31-21 with 3 seconds left when Andy Dalton threw an interception on the game’s final play.
Trailing 30-14 against the Chiefs, with 9 seconds left, Blake Bortles threw an interception on the game’s second-to-last play.
All of these are examples of meaningless interceptions, at least from a win probability perspective. These teams all had a less than one percent chance of winning the game, and the interception therefore was not meaningful. Tom Brady had one of these, too: the Patriots had a win probability of over 99% when, up 38-0 in the 4th quarter, he threw an interception. That pick was meaningless.
We should not treat all interceptions equally, and with win probability data becoming more mainstream, we are no longer restricted by a game’s box score. A special thanks to Ron Yurko (@Stat_Ron) for providing the raw data.
There have been 190 interceptions in the NFL through 7 weeks. Of those, 120 of them reduced the offense’s win probability by 10% or less, while only 17 dropped the team’s win probability by 20% or more.
The table below shows each quarterback, his number of interceptions this season, the total win probability he cost his team on those interceptions, and the average drop in win probability per interception.
| Quarterback | INTs | Total WP Change | Avg WP Change |
|---|---|---|---|
| Darnold | 10 | -100% | -10% |
| Keenum | 9 | -77% | -9% |
| Carr | 8 | -100% | -12% |
| Luck | 8 | -80% | -10% |
| Dalton | 8 | -68% | -9% |
| Bortles | 8 | -41% | -5% |
| Beathard | 7 | -106% | -15% |
| Watson | 7 | -86% | -12% |
| Brady | 7 | -30% | -4% |
| Roethlisberger | 6 | -62% | -10% |
| Trubisky | 6 | -53% | -9% |
| Winston | 6 | -37% | -6% |
| Mariota | 5 | -73% | -15% |
| Rosen | 5 | -57% | -11% |
| Tannehill | 5 | -51% | -10% |
| Fitzpatrick | 5 | -45% | -9% |
| Mahomes | 5 | -36% | -7% |
| Mayfield | 5 | -36% | -7% |
| Goff | 5 | -33% | -7% |
| Stafford | 5 | -31% | -6% |
| Allen | 5 | -30% | -6% |
| Manning | 4 | -59% | -15% |
| Peterman | 4 | -51% | -13% |
| Prescott | 4 | -41% | -10% |
| Flacco | 4 | -35% | -9% |
| Newton | 4 | -34% | -9% |
| Wilson | 4 | -29% | -7% |
| Bradford | 4 | -12% | -3% |
| Cousins | 3 | -32% | -11% |
| Garoppolo | 3 | -22% | -7% |
| Rivers | 3 | -20% | -7% |
| Anderson | 3 | -4% | -1% |
| Taylor | 2 | -70% | -35% |
| Osweiler | 2 | -22% | -11% |
| Smith | 2 | -13% | -7% |
| Ryan | 2 | -11% | -6% |
| Gabbert | 1 | -20% | -20% |
| Foles | 1 | -18% | -18% |
| Wentz | 1 | -15% | -15% |
| Kizer | 1 | -9% | -9% |
| Rodgers | 1 | -9% | -9% |
| Kessler | 1 | -4% | -4% |
| Cassel | 1 | 0% | 0% |
Tyrod Taylor has thrown two interceptions this year, but they’ve been two really meaningful interceptions. Tom Brady and Blake Bortles have thrown 7 and 8 interceptions, respectively, this year, and their combined 15 interceptions have cost their teams about the same amount of win probability as Taylor’s two picks.
For Brady, 4 of his 7 interceptions came with the Patriots having a win probability of 94% or greater: up 38-0 against Miami, up 24-10 twice in the second half in Colts territory, and up 38-24 with 6:31 left in the game in Bears territory. Interceptions are bad, but none of them were particularly meaningful.
Brady also threw a pick against the Lions when New England’s win probability was only 10%.
His other two interceptions came when the game was more in doubt, both on third down; one against the Texans and one against the Dolphins.
All interceptions are bad, but some are very meaningful and some are practically meaningless. Brady’s seven interceptions, in the aggregate, aren’t as bad as the one Nathan Peterman interception against the Texans. With 1:30 left in the 4th quarter in a tie game, Peterman threw a pick six that cost the Bills 46% in win probability.