QB Annual: 5 Takeaways from PFF's Interception Luck Metric

Five Takeaways from PFF's Interception Luck Study

Interception totals are one of the most heavily weighted statistics in quarterback evaluation. They shape MVP debates, influence contract negotiations and follow players through their careers in a way that only one other number does. But a decade of PFF grading tells a more complicated story: a significant portion of every quarterback's interception total has little to do with the quarterback himself.

To measure the gap, PFF built a metric called Net Luck, one of several new additions coming to this summer's edition of the PFF QB Annual. It works from a simple premise: not all interceptions are created equal. Some are thrown on passes so badly placed that a turnover was the expected outcome. Others land in a defender's hands on throws the quarterback executed perfectly well, through no fault of their own. Net Luck asks how many interceptions a quarterback should have thrown based on the quality of their decisions, then compares that number to how many they actually did.

From a sample of 209 consecutive-season quarterback pairs since 2016, roughly 88% of a quarterback's deviation from league average disappeared the following season. In other words, if a quarterback was unusually lucky or unlucky one year, history suggests that probably won't continue the next.

Here are some takeaways from the study and a snippet of what you'll be able to see in the 2026 QB Annual.

1️⃣ Matthew Stafford has been the luckiest QB in football for two years running

Stafford's 2024 season was the single luckiest quarterback season in the 10-year dataset. Based on the quality and volume of his turnover-worthy throws, the model estimates that he finished with roughly 7.3 fewer interceptions than an average quarterback would have been expected to throw in the same circumstances. His 2025 season ranks third all-time by the same measure, at 6.8 fewer interceptions than expected.

Those results weren't driven by a single source. Over that two-year stretch, defenders dropped 22 of Stafford's would-be interceptions — 13 in 2024 and nine in 2025 — while he also benefited from a lower-than-average rate of non-turnover-worthy throws becoming interceptions. Together, those factors helped produce interception totals that looked significantly cleaner than the tape would suggest.

The trend extends beyond the past two seasons. Stafford's Net Luck score of +17.1 since 2016 means he has finished with roughly 17 fewer interceptions than expected over that stretch, trailing only Aaron Rodgers (+24.2) in the dataset. His 84 dropped interceptions are also the most of any quarterback during that span.

None of this diminishes the quality of Stafford's play. The most important skill remains avoiding turnover-worthy throws in the first place, and Stafford has generally done that well throughout his career. Net Luck simply separates that skill from the resulting variance. Over the past two seasons, that variance has broken heavily in Stafford's favor, and history suggests a third consecutive season at that level would be highly unusual.

2️⃣ Geno Smith‘s 17-interception season was not as bad as it looked

Smith finished 2025 as the unluckiest quarterback in football. Based on the quality and volume of his turnover-worthy throws, the model estimates that he finished with roughly 4.7 more interceptions than an average quarterback would have been expected to throw in the same circumstances.

The final box score credited Smith with 17 interceptions, the highest total in the NFL. The grading paints a different picture. He recorded 18 turnover-worthy throws, a manageable figure for a full-time starter, but 12 of those plays became interceptions. That 66.7% conversion rate sat nearly 18 percentage points above the league average. He also threw five interceptions on passes that were not deemed turnover-worthy, well above the roughly three to four interceptions his attempt volume would typically produce.

Under league-average interception luck, Smith's season would have looked much different. The model projects that his underlying performance was closer to a 12- or 13-interception campaign than the 17-interception season reflected in the box score.

3️⃣ Josh Allen owns the two most extreme year-to-year swings in the dataset — in consecutive seasons

If there is a quarterback who demonstrates the volatility of interception outcomes, it's Allen.

In 2023, 76.5% of Allen's turnover-worthy throws became interceptions, one of the three highest rates in the 10-year dataset. A year later, that figure plummeted to 16.7%, the second-largest year-over-year drop on record. Allen went on to win NFL MVP. In 2025, the rate jumped back to 75.0%, the third-largest swing in the opposite direction.

The remarkable part is how little Allen himself changed. Across all three seasons, his turnover-worthy throw total remained relatively stable, ranging from 12 to 17.

In practical terms, Allen's interception count swung dramatically from season to season despite his level of turnover-worthy play remaining fairly consistent. Defenders simply converted those opportunities into interceptions at wildly different rates from one year to the next.

That three-year stretch is one of the clearest examples of why interception totals can be misleading when viewed in isolation. Allen's 2023, 2024 and 2025 seasons produced vastly different turnover outcomes despite relatively similar underlying play. It's also a reminder that an MVP season and a disappointing statistical follow-up can sometimes be separated more by variance than by any meaningful change in the quarterback himself.

4️⃣ Brock Purdy has been the unluckiest quarterback in football over the past four seasons

Purdy's career Net Luck figure of -7.7 ranks fifth-worst among quarterbacks with at least 1,500 attempts since 2016. In simple terms, the model estimates that he has thrown roughly eight more interceptions than an average quarterback would have been expected to throw in the same situations.

What's unusual is that the trend has persisted. Purdy's career turnover-worthy-throw-to-interception rate sits at 65.1%, the highest among qualifying quarterbacks and more than 16 percentage points above the league average. Most quarterbacks see that type of number regress toward the mean over time. Purdy has yet to.

5️⃣ Not all interception totals are created equal

Quarterbacks can arrive at similar interception totals through completely different processes, and the 2025 season provided several extreme examples.

Dak Prescott benefited from more dropped interceptions than any quarterback in football. Defenders dropped 11 potential interceptions against Prescott, two more than any other passer despite him attempting 116 fewer passes than the next-closest quarterback. Seven of those drops came on turnover-worthy throws, helping hold his turnover-worthy-throw-to-interception conversion rate to just 35.3%.

Michael Penix Jr. arrived at a similarly favorable outcome through a different route. He threw just three interceptions despite recording nine turnover-worthy throws, with defenders dropping four of those opportunities. He also avoided the random interceptions that often inflate quarterback totals, throwing zero interceptions on non-turnover-worthy passes across 276 attempts.

Justin Herbert experienced the opposite. Five of his 12 turnover-worthy throws became interceptions, but the bigger issue was what happened on his better throws. Herbert led the NFL with eight interceptions on non-turnover-worthy passes, including one that came after a receiver dropped a catchable pass into a defender's hands. He finished with 13 interceptions despite an expected total closer to 10.

The takeaway is simple: interception totals often describe outcomes, not process. Two quarterbacks can finish with similar numbers in the box score while taking dramatically different paths to get there.