Adjusted Interceptions 2019
This is an article about interceptions in 2019, and that means we are going to talk a lot about Jameis Winston. Jameis Winston shall be the player most frequently discussed today, and the player most frequently discussed today shall be Jameis Winston. Jimmy Garoppolo and Joe Flacco will not come up, excepting that they are compared to Jameis Winston. Jameis Winston, Jameis Winston, Jameis Winston. Jameis, Jameis, Jameis. Winston, Winston, Winston. Jameis "Jameis Winston" Winston. Jameis Winston. Wameis Jinston. Jameston Winis. Jameis Winston.
Specifically today we are going to discuss adjusted interceptions. Unlike the NFL's raw interception totals, these numbers account for plays when a defender drops a pass that he should have caught, or when a wide receiver makes a big play to turn what should have been a turnover into an incompletion instead. On the other hand, sometimes quarterbacks are charged with interceptions that aren't really their fault -- passes that bounce off a receiver's hands and straight to a defender -- or interceptions that don't matter, like Hail Mary passes.
After each season, we go back and account for these discrepancies and account for each quarterback's adjusted interceptions. Here's the process:
- We start with each player's actual interception total. Jameis Winston led the NFL in 2019 with 30 interceptions, only the 12th time in NFL/AFL history that a quarterback has thrown so many in a single season, and the first since Vinny Testaverde in 1988.
- We then add plays where the quarterback threw a ball that could have or should have been intercepted but was not, either because the defender outright dropped the ball (which we have been tracking in game charting since 2007), or he had it knocked out of his hands by an offensive receiver (a "defensed interception," which we have been tracking since 2012). These are listed as "Drop/Def INT" in the table at the end of this page. Winston also led the NFL in this category, with 13 -- the most of any quarterback in a single season since Andrew Luck threw 14 as a rookie in 2012. (Note that this does not include a Kyle Allen pass that was dropped by Adrian Amos of the Packers, only to then be intercepted by Amos' teammate Tramon Williams -- Allen deserved to be intercepted on that play, and eventually, he was.)
- Next, we subtract those interceptions that were tipped by receivers into the hands of defenders (as established in 2017, these plays can be thought of as Matt Ryan specials). Three players tied for the league lead in this category with three apiece: Jimmy Garoppolo, Ryan Fitzpatrick, and Baker Mayfield. (Jameis Winston had two.) We also subtract passes that are tipped by receivers but then dropped by defenders to make sure they are not double-counted. There was only one of these in 2018, but eight in 2019, though no player threw more than one.
- We subtract Hail Mary interceptions, as well as interceptions thrown in desperation on fourth down in the final two minutes of a game. We're flexible on these definitions, but this year there was only one play that was in the gray area: a Winston interception in Week 6 against Carolina, thrown on fourth down while trailing by 11 points with 2:31 to go. We opted not to count that one as a Hail Mary interception. Three players threw two qualifying Hail Mary interceptions last season: Dak Prescott, Philip Rivers, and Detroit's David Blough. (Blough did not qualify for the final table below, but he finished with eight adjusted interceptions and six actual interceptions.)
- New this year, we are subtracting dropped interceptions that occur in Hail Mary situations, since those plays wouldn't count as adjusted interceptions even if they had been caught. Mitchell Trubisky had two such plays; Josh Allen, Daniel Jones, and Aaron Rodgers threw one each.
When we started running these numbers, we had to get the data from our own in-house volunteer game charters. For the last five seasons, we've had access to data from Sports Info Solutions. Determining whether or not a defender should be charged with a dropped interception will always be subjective on some plays, but you can rest assured that all the obvious calls have been counted here.
Anyway. The 2019 leader in adjusted interceptions was -- of course -- Jameis Winston.
|Adjusted Interceptions, 2019|
We have adjusted interception numbers dating back to 2007, and Winston is the first player since then to hit 40 adjusted interceptions in a season. Only three other players have even hit 30 in a season: Jay Cutler (34 in his first year with the Bears in 2009), Eli Manning (31 in 2013), and Luck (30 in 2012).
The gap between Winston and anyone else last year was monstrous. Philip Rivers and Kyle Allen each threw 26 adjusted interceptions. That was enough to tie for second place, but they were just as close to the 12 adjusted interceptions of 18th-place Josh Allen as they were to Winston. Baker Mayfield (23) and Andy Dalton (21) round out the top five.
In Winston's defense, he threw a lot of passes into the hands of defenders in part because he threw a lot of passes period, a league-high 624. He threw an adjusted interception on 6.4% of his passes. That's still the highest rate in several years, but at least it's not totally unprecedented. The last qualifying passer with a higher rate was Chicago's Matt Barkley in 2016. The highest rate on record is the 7.1% of Carson Palmer, who threw 23 adjusted interceptions in 326 passes for the Oakland Raiders in 2011.
Because he threw so many interceptions, Winston is going from the starting quarterback for the Buccaneers to the backup for the Saints. That's going to be a radical shift in New Orleans, where last year's backup, Teddy Bridgewater, had the lowest adjusted interception rate in the league. Bridgewater, meanwhile, is moving on to Carolina, where he will replace Kyle Allen … whose adjusted interception rate of 5.3% was second-highest behind Winston.
With all those hundreds of passes, Winston's luck mostly evened out. The average quarterback throws about 30% more adjusted interceptions than actual interceptions. At that rate, we would expect a passer with 40 adjusted interceptions to finish with 30.8 actual interceptions, barely any higher than Winston's real-life total of 30. Compare that to Philadelphia's Carson Wentz, who was 12th in the NFL with 16 adjusted interceptions but tied for 20th with seven real-life picks. That's 5.3 interceptions less than we would expect, making Wentz the luckiest quarterback in the league. He's followed in that department by Kyle Allen (-4.0), Drew Brees (-3.7), Tom Brady (-2.8), and Patrick Mahomes (-2.7).
The unluckiest quarterback was Cleveland's Baker Mayfield. Mayfield threw 23 adjusted interceptions, so we would expect him to have thrown 17.7 total interceptions, but instead he threw 21. That difference of +3.3 was the highest in the league. It's even higher than the +3.2 difference Mayfield had in 2018, when he was second in bad luck behind Jacksonville's Blake Bortles. In two NFL seasons, the defenders of the NFL have intercepted 35 of Mayfield's passes, but dropped only eight of them. Last season, he was joined in misfortune by Pittsburgh's Devlin Hodges. Hodges only threw 160 passes, so you won't find him in the above table, but he was one of the few quarterbacks to throw more actual interceptions than adjusted interceptions, with eight of the former and seven of the latter -- 2.6 more interceptions than expected. Mayfield and Hodges are followed by Dwayne Haskins (+2.4), Ryan Tannehill (+2.2), and Hodges' teammate Mason Rudolph (+2.1). Add in Ben Roethlisberger's numbers (one interception, two adjusted interceptions), and Steelers quarterbacks threw 4.2 more interceptions than expected. It truly was a cursed year in Pittsburgh.
27 comments, Last at 08 May 2020, 12:05pm
#1 by theslothook // May 04, 2020 - 11:15am
As always, everything in football is context dependent. I don't think it's an accident that the coach of Luck in 2012( when he threw a ton of ints) is the same coach of Jameis. A different coach probably alters the game plan or tries to at least mitigate Jameis' worst tendencies. But Bruce's offenses have never been gun shy. There's something admirable in that.
#4 by Roch Bear // May 04, 2020 - 1:55pm
Interesting indeed. I wonder what the correlation of adj. int% would be if X-1, X-2, X+1, and X+2 were averaged together for comparison. And what I'm suggesting here is one of the original sins of commentary, 'hey why doesn't somebody do this thing I'm too lazy to do.'
#5 by Aaron Schatz // May 04, 2020 - 2:06pm
All correlations regarding interceptions are fairly weak. It's stronger than the correlation between actual interception rate and Y+1 interception rate, or at least it has been in the past.
I also wrote a bit about using passes defensed to predict interceptions at midseason for ESPN+. Again, low correlations, but better than just using standard interception totals to predict future interception totals.
#12 by BigRichie // May 04, 2020 - 6:05pm
The weaker the correlation, the less reason to believe it's actually measuring what it's supposed to.
Predicting 'Brady will % throw more interceptions and Winston fewer' is about as low as fruit can hang. But if Brady goes way up and Winston way, way down, well ...
#21 by Pat // May 05, 2020 - 11:28am
What are you talking about? That's actually pretty strong.
First, you'd never expect it to be high. You're comparing an intrinsically noisy statistic to another noisy statistic. Interceptions are a counting statistic, so they fundamentally have around ~20-30% noise, just because you're only talking about 10-15 events/year. Plus interception opportunities increase when trailing, but trailing opportunities from year-to-year are also effectively going to be random (with constant team strength) since you don't control your schedule.
There's also the "you don't keep playing crappy quarterbacks" problem, so the high "next year actual int" values are going to have the high "current adjusted int" values missing, biasing their results down.
Fundamentally you can't ever beat the problem of low statistics, so you're never going to get a solid measure on what you're trying to get at (quarterback inaccuracy/bad decision rate). You're just going to eke out incremental improvements.
#22 by takeleavebelieve // May 05, 2020 - 2:37pm
I’m not sure I understand the point you’re trying to make. Have you tested other metrics and found one with r-sq > .16?
Also, keep in mind that Adjusted INTs aren’t necessarily meant to be predictive; they’re meant to provide some additional insight to a counting stat that’s very context-dependent.
#10 by Dave Bernreuther // May 04, 2020 - 4:05pm
I'm curious about the difference between the charting here (13) and the charting&video Cian made here:
Obviously some are tougher catches than others, but the ones highlighted in that video are definitely not of the "required a superhuman effort to possibly intercept" variety.
The ones I can see being judgment calls here are:
- vs Saints, ~ :31 (out of bounds even though only the defender could've caught it)
- vs Seahawks, next play (can't tell what actually happened on the receiving end; pretty damn good pocket movement before that actually)
- vs Titans, ~ .47 (laid up short)
- vs 49ers, 1:02 (this looks more like a defensed pass to me, though he could easily have gotten both hands on it)
There aren't all 21 in that video either, now that I track it.
Doesn't make me any less curious, though. It's been a lot of fun to point to the 30+21 as a reason to mock his supporters. 40 is still a ton, but it's really just not anywhere near as fun as 50+.
Either way, it makes me laugh that anyone would have ever suggested that Belichick may have been interested in him. The whole reason they won with Brady early on as he developed was that he was risk averse and avoided turnovers in a way that even a non-Arians coached Jameis could never dream of. And that's not something you coach out of a guy in his sixth season.
#11 by Vincent Verhei // May 04, 2020 - 4:25pm
Without going over things one play at a time, I believe Cian charts "interceptable passes" -- i.e., passes that COULD have been intercepted. The folks at SIS chart dropped interceptions -- passes that SHOULD have been intercepted. That would explain the discrepancy, and why Cian's numbers are higher.
#16 by RevBackjoy // May 04, 2020 - 11:28pm
Next Gen stats could shed a lot of light in this situation- every pass, really, has an interceptability (i.e. probability of getting picked off) between 0% and 100%, so each pass contributes between 0 and 1 "expected interceptions". The threshold of interceptable vs not is entirely arbitrary. If you set the limit at 50%, a 51% pass would be "interceptable", while a 49% would not. A better way to think of it would be: Throw 1 = 10% INT chance, Throw 2 = 60% INT chance, Throw 3 = 40%, for 1.1 expected INTs.
To develop a metric like this would take a lot of thought and research (and hence $), but given that we already have an expected completion percentage stat (and, for MLB, expected hit probability of balls in play), I'm sure it could be done. At the end of a game/season, you could just add the Expected Interception Prob of each pass of each QB, which would yield the total # of expected picks.
#15 by mehllageman56 // May 04, 2020 - 9:43pm
Going by Brady's interception percentage by year, he was less risk averse prior to 2007, the Randy Moss year; every year he started it's 2.3 and above. Only once after 2007 did Brady's interception percentage rise above 2.0, in 2009 (2.3). While I agree Brady was smarter with the ball than Winston back then, I wouldn't argue that's Brady defining strength; I think his pocket presence was second to none in NFL history for most of Brady's career. I also think risk aversion can become a problem if it costs your team points; I'd rather have Brady than Tyrod Taylor or peak Ken O'Brien (in his good years, O'Brien interception rate would lead the league). You have to take some risks, just not all the ones Jameis goes for.
#17 by RevBackjoy // May 04, 2020 - 11:34pm
My favorite is pass #4, at about 0:16, against the Rams. His receiver has beaten his defender by multiple steps and is streaking into the end zone. Yet the ball lands about 5-10 yards short and several yards to the left, leading to a three-way collision between two Rams defenders and a (different) Bucs receiver! Does that count as "interceptable"?
#23 by ammek // May 06, 2020 - 6:46am
When you published the multi-season figures in 2015, Drew Brees came out as the unluckiest pickee in the previous decade, with 12.2 interceptions more than expected. With his luck having swung in the opposite direction in 2019, is Brees still number one in that list?
At the other end, Mahomes is now 6.9 INTs below expected through two seasons as a starter. Is there any reason to believe that good fortune will continue?
#24 by Eddo // May 06, 2020 - 9:08am
This doesn't directly answer your question about Brees, but it does seem to indicate his very low interception rates the last three years is a bit "lucky".
re: Mahomes, I wonder if arm strength correlates with "luckiness" with regards to interception rate. It stands to reason that defenders would have a more difficult time catching passes that were thrown faster. Of course, that assumes a lot of things about the dropped interceptions for Mahomes.
#25 by dank067 // May 06, 2020 - 10:44am
I wonder if FO or anyone else who charts dropped INTs like this has looked to see if there's any relationship between drops and depth of target. The Saints offense over the past several seasons has really concentrated its throws to within 10 yards of the line of scrimmage. Without as much time to react to a pass or track its trajectory in the air, are defenders more likely to drop an interception opportunity?
#26 by Joseph // May 07, 2020 - 1:25pm
One could also argue that Michael Thomas went from "pretty good #2 receiver" to All-Pro status, and Alvin Kamara went from the "University of Tennessee bench" to "Pro Bowl caliber, hard to match up with RB." These 3 events might be somewhat correlated.
#27 by horn // May 08, 2020 - 12:05pm
Wentz got 'lucky' but maybe some of those 'should have been INTs' were throws he howitzered into tight windows [since he had no actual NFL WRs to work with] that most LBs/S aren't ever going to be good at catching.