Writers of Pro Football Prospectus 2008

08 Apr 2013

Adjusted Interceptions 2012

by Aaron Schatz

Two years ago, we introduced a new metric on FO called "adjusted interceptions." The basic idea:

  • We add in plays where the quarterback only escaped an interception because the defender couldn't hold onto the ball (dropped interceptions, which we've been tracking in game charting since 2007).
  • We subtract plays where the interception is tipped to the defender by a receiver who should have caught the pass.
  • We subtract Hail Mary interceptions as well as interceptions thrown in desperation on fourth down in the final 2:00 of a game.

Obviously, there's going to be a little bit of subjectivity here, but our game charters do their best. If the defender has to dive for a ball only to have it bounce off his fingertips, that's not a dropped interception.

This year, we made one change to how we charted interceptions and near-interceptions. Unlike in past years, we also charted passes that would qualify as "dropped interceptions" except that the defender's catch was broken up by the receiver. These "defensed interceptions" count against the quarterback but will not be included in the dropped interception count for defenders.

Interceptions are notoriously hard to forecast from year to year, because there's so much random chance and statistical noise involved. However, we did find that this new metric, adjusted interceptions, had a higher year-to-year correlation than standard interception totals and was a better predictor of future interceptions. But something very strange happened in 2012. The year-to-year correlation of adjusted interception rate was very high (.37), but the year-to-year correlation of actual interception rate was even higher (.52). I'm guessing this is a one-year fluke, and in the long run, adjusted interceptions are still a better predictor of the future than standard interception counts.

Andrew Luck led the league in dropped interceptions in 2012; in fact, dropped interceptions show that the efficiency gap between Luck and fellow rookies Robert Griffin and Russell Wilson was even larger than shown by stats like DVOA. Obviously, this is partly due to the fact that Luck was throwing the ball deeper and in harder situations than Griffin or Wilson, but the difference is still dramatic. Luck had 18 interceptions, the second-highest total in the NFL, and also had 14 dropped picks. Wilson had just 10 picks and five dropped picks, while Griffin had five of each. Luck's total of 14 dropped interceptions is the second-highest season total since 2007, behind only Mark Sanchez's 15 for 2010.

No quarterback threw more than one Hail Mary interception this year, but Chad Henne did throw three fairly meaningless interceptions: a Hail Mary at the end of the Week 16 game against New England, a pick on fourth-and-1 with 1:59 left and the Jaguars losing to the Colts by 17 in Week 10, and a first-down pick that came with 0:38 left and the Jaguars losing to the Bills by 16 in Week 13.

Alex Smith once again did a fabulous job of taking care of the ball, at least until losing his job. He had no dropped interceptions. Neither did Sam Bradford, even though he had 13 actual interceptions. Other quarterbacks who were underrated by conventional interception totals included Brady Quinn, Matt Cassel, Blaine Gabbert, and Kevin Kolb.

Matt Hasselbeck and Nick Foles actually managed to have more dropped interceptions than they did actual interceptions. Other quarterbacks who were overrated by conventional interception totals included Luck, Brandon Weeden, and Eli Manning.

The following table lists adjusted interceptions for all quarterbacks with at least 200 pass attempts. Note that the average quarterback's adjusted interception rate is 0.8 percent higher than his actual interception rate.

Player Team INT HM/
End Q4
Drop/
Def INT
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate
14-R.Lindley ARI 7 0 2 0 9 171 4.1% 5.3%
3-B.Weeden CLE 16 0 10 0 26 517 3.1% 5.0%
19-J.Skelton ARI 9 1 2 0 10 200 4.5% 5.0%
8-M.Hasselbeck TEN 5 0 6 0 11 221 2.3% 5.0%
12-A.Luck IND 18 1 14 1 30 620 2.9% 4.8%
6-M.Sanchez NYJ 18 0 5 2 21 452 4.0% 4.6%
6-J.Cutler CHI 14 0 6 0 20 435 3.2% 4.6%
7-C.Henne JAC 11 3 6 0 14 305 3.6% 4.6%
10-E.Manning NYG 15 0 9 0 24 536 2.8% 4.5%
5-J.Freeman TB 17 0 8 1 24 556 3.1% 4.3%
14-R.Fitzpatrick BUF 16 0 5 0 21 503 3.2% 4.2%
9-N.Foles PHI 5 0 8 2 11 264 1.9% 4.2%
10-J.Locker TEN 11 1 4 1 13 315 3.5% 4.1%
Player Team INT HM/
End Q4
Drop/
Def INT
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate
7-M.Vick PHI 10 0 4 0 14 349 2.9% 4.0%
9-T.Romo DAL 19 1 8 1 25 646 2.9% 3.9%
7-C.Ponder MIN 12 0 7 0 19 484 2.5% 3.9%
7-M.Cassel KC 12 1 2 3 10 270 4.4% 3.7%
14-A.Dalton CIN 16 1 5 1 19 525 3.0% 3.6%
3-C.Palmer OAK 14 0 8 2 20 564 2.5% 3.5%
9-D.Brees NO 19 1 8 3 23 667 2.8% 3.4%
17-P.Rivers SD 15 1 4 0 18 526 2.9% 3.4%
17-R.Tannehill MIA 13 0 3 0 16 482 2.7% 3.3%
5-J.Flacco BAL 10 0 7* 0 17 531 1.9% 3.2%
9-M.Stafford DET 17 0 7 1 23 723 2.4% 3.2%
1-C.Newton CAR 12 0 3 0 15 484 2.5% 3.1%
9-B.Quinn KC 8 1 0 1 6 198 4.0% 3.0%
Player Team INT HM/
End Q4
Drop/
Def INT
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate
3-R.Wilson SEA 10 1 5 2 12 397 2.5% 3.0%
7-B.Roethlisberger PIT 8 0 5 0 13 443 1.8% 2.9%
8-M.Schaub HOU 12 0 3* 0 15 544 2.2% 2.8%
7-C.Kaepernick SF 3 0 3 0 6 218 1.4% 2.8%
2-M.Ryan ATL 14 0 3* 1 16 615 2.3% 2.6%
10-R.Griffin WAS 5 0 5 0 10 395 1.3% 2.5%
18-P.Manning DEN 11 0 3 0 14 581 1.9% 2.4%
11-A.Smith SF 5 0 0 0 5 219 2.3% 2.3%
8-S.Bradford STL 13 1 0 0 12 544 2.4% 2.2%
11-B.Gabbert JAC 6 0 1 1 6 278 2.2% 2.2%
12-A.Rodgers GB 8 0 4 1 11 551 1.5% 2.0%
4-K.Kolb ARI 3 0 0 0 3 183 1.6% 1.6%
12-T.Brady NE 8 0 5* 3 10 634 1.3% 1.6%
*Does not include one pass that was dropped/tipped by the receiver and then also dropped by a defender.

Posted by: Aaron Schatz on 08 Apr 2013

34 comments, Last at 14 Apr 2013, 11:58pm by NeShep

Comments

1
by mavajo (not verified) :: Mon, 04/08/2013 - 2:26pm

Wilson has 1 HM/End Q4. Should be 2. 9/24/13 - Never Forget.

2
by Hughughugh (not verified) :: Mon, 04/08/2013 - 2:47pm

Tate's feet landed first with both hands on the ball, which is required for actual possession.

At least in the real world.

6
by Anonymous485 (not verified) :: Mon, 04/08/2013 - 4:27pm

You must be a Seahawks fan. The rest of the people who watched the game were bewildered by the terrible call.

7
by Anonymous6546 (not verified) :: Mon, 04/08/2013 - 4:49pm

The situation was unfortunate, but I don't think it's accurate to call the call "terrible". Lot's of people thought it was, but they're wrong. There's no such thing as "more" or "better" possession. It was a bang bang play, and a tie goes to the offense. I wouldn't want my team to lose that way either, but if that wasn't an example of when simultaneous possession is the correct call, I don't know when would be. Of course, they also should have called the offensive PI, which though they never do, was particularly egregious. (But wouldn't have meant another INT for Wilson ;))

9
by Anonymous6546 (not verified) :: Mon, 04/08/2013 - 4:51pm

I should have added this above, as it sounds like I think they definitely got the call right, and I'm not saying that either. The call, in my mind, was correct at best to understandable at worst.

8
by Anonymous78234123412341234123412341912091 (not verified) :: Mon, 04/08/2013 - 4:49pm

I'm a Patriots fan and I think it was the right call...the rules say it was so it was

12
by Anonymousover9000 (not verified) :: Mon, 04/08/2013 - 7:23pm

"Patriots Fan". Void opinion.

13
by Insancipitory :: Mon, 04/08/2013 - 8:22pm

Not the supervising officials who weren't on strike, and not the league office people who said the catch was properly officiated. Sure, they admitted the no-call on the OPI was a mistake, but it's not like Seattle would be getting all the uncalled Packer holding back. Green Bay fans should be thankful for the replacement refs for their merciful interpretation the Packers line play, Rodgers lived to play again.

Tie goes to the runner. :)

15
by Anonymous!! (not verified) :: Mon, 04/08/2013 - 9:15pm

Is this where the people mention that the replay official, the official that signed off on the call, was a regular official. I really think people loved to place the blame on the replacement officials, when it was a call agreed to by a regular (non-replacement) official in the booth.

17
by Anonymous Dude (not verified) :: Tue, 04/09/2013 - 1:10am

I absolutely love, LOVE, that we are still arguing about this. God bless the NFL.

23
by Kevin o (not verified) :: Tue, 04/09/2013 - 10:28am

Well, since the call could not be overturned once it was made on the field (nonreviewable), your argument is invalid.

28
by Wikitorix (not verified) :: Tue, 04/09/2013 - 3:19pm

If it was a nonreviewable play, then the biggest error the refs made was reviewing it. The play was reviewed, and the call stood as called.

I'm pretty sure that the play was reviewable, though, since all touchdowns are automatically reviewed.

There's video evidence of this - you don't get to just make up stuff.

31
by Anonymous/(); (not verified) :: Thu, 04/11/2013 - 4:21pm

Wrong thread guys. Last minute Hail Mary s don't show up in this metric.

32
by Steve McQueen (not verified) :: Sat, 04/13/2013 - 7:39pm

Yeah, no kidding. Seems like no one bothered to read what goes into the metric.

27
by BigWoody (not verified) :: Tue, 04/09/2013 - 2:07pm

mavajo,

If you will draft a letter to the league office requesting the MNF game be awarded to the Packers, I as a Seahawk fan, will co-sign it. However, the letter must also include the requirement that the Lombardi Trophy for Super Bowl XL be taken from the Steelers and is awarded to the Seahawks. - Never Forget

3
by cjfarls :: Mon, 04/08/2013 - 2:56pm

Interesting... any thoughts from Colts fans, etc. on how negative that INT rate portends for Luck? I know he needs a lot of allowance for being dropped right into the fire of a full-fledged NFL passing attack as a rookie, and that rookie INT rates can fall precipitously...

Anyway, does this look like normal rookie getting up to speed growing pains, or should we all be wondering if his decision-making/anticipation/accuracy could be potential career killing issues? I haven't heard such concerns nor notice it in the one Colts game I saw, but adding 14 more ints to his already high 18 certainly gives one pause....

4
by Led :: Mon, 04/08/2013 - 3:24pm

I'd take that 14 dropped interception number with a grain of salt. With a subjective measure like dropped interceptions that's difficult to quality control, you should be careful about the extreme outliers. I'm sure it's right directionally -- Luck was probably somewhat, err, lucky. But was he really worse than Mark Sanchez and comparable to QB Arizona at taking care of the ball? I'd also question Bradford's zero dropped picks in 544 attempts for the same reason.

33
by Steve McQueen (not verified) :: Sat, 04/13/2013 - 7:41pm

It's hard to say if he was worse, as those interceptions were dropped. If a normal amount of them were caught people maybe would be comparing him to some of those QBs you listed (with the rookie caveat).

14
by Anonymouse (not verified) :: Mon, 04/08/2013 - 8:35pm

Not a Colts' fan, but it worked out well enough last time...

Peyton Manning led the league in attempts and interceptions as a rookie (575 attempts, 28 picks).

http://pfref.com/tiny/LMnJF

Manning's interception rate dropped precipitously thereafter, to a number closer to his career average of 2.7%.

http://www.pro-football-reference.com/players/M/MannPe00.htm

Luck also threw a lot of passes (617) with an actual interception percentage of 2.9%, which was a bit more than the league average. That doesn't take into account the adjustments in the above article, but I don't know if we can draw any meaningful inferences regarding how well a QB's rookie interception total (adjusted or otherwise) relates to his future turnover performance.

16
by Sdgmen (not verified) :: Tue, 04/09/2013 - 12:36am

I like the stat, certainly provides additional analysis and food for thought in ranking/rating QB peqrformance. The only issue I see is that if dropped interceptions are considered, where do interceptions due to receivers falling down or running incorrect routes comes into play? As Giants fan I am we'll aware of the need for receivers and QB's to be on the same page in regard to their reads. This is especially critical on "hot reads". If the receiver cuts the wrong way, many times a quarterback can end up throwing the ball right to a defender.

21
by Ryan D. :: Tue, 04/09/2013 - 8:14am

All I can see is Steve Smith slipping down on the turf in Chicago, and the resulting gifted interception being returned for a touchdown.

24
by Aaron Schatz :: Tue, 04/09/2013 - 10:40am

It's hard to judge "wrong route." That's much more subjective, even though it looks like it is responsible for pretty much all interceptions on passes from Tony Romo intended for Dez Bryant. No matter what some readers might think, "dropped interception" really isn't that subjective, and when it is, we always err on the side of not penalizing the QB for that pass.

If you are curious, 16 different picks are marked "miscommunication" for reason. Two players had more than one, surprisingly neither of them is Romo (again, an example of how subjective the concept is). Carson Palmer had two in one game (Week 9 vs. TB) and Peyton Manning had "miscommunication" interceptions in Weeks 6 and 16. Eli Manning actually didn't have any.

I guess "receiver tripped" is more valid to remove from adjusted interceptions, though. The following players had an interception marked "receiver tripped":

Carson Palmer Week 3 to Denarious Moore
Ryan Tannehill Week 4 to Brian Hartline
Russell Wilson Week 4 to Anthony McCoy
Cam Newton Week 8 to Steve Smith
Matt Schaub Week 10 to Keshawn Martin
Jay Cutler Week 14 to Alshon Jeffery
Josh Freeman Week 16 to Mike Williams

26
by nat :: Tue, 04/09/2013 - 11:29am

No matter what some readers might think, "dropped interception" really isn't that subjective, and when it is, we always err on the side of not penalizing the QB for that pass.

It would depend on the distribution of catch difficulty, wouldn't it?

Do passes that touch a defender's hands really group into "easy interception" and "incredible play needed to intercept" without a smooth transition in the middle ranges? That seems very unlikely, unless you are using some bright-line distinction as a proxy for difficulty.

19
by bubqr :: Tue, 04/09/2013 - 3:30am

Everyone is clearly in awe of Andrew Luck, and while I see the talent, leadership, etc. I'm not as 100% sold on him as mostly everyone else.
Everyone agrees that he's already a lock for the Hall of Fame (I don't think that's an exaggeration), and I'm still not that confident after having seen 3 of his games last year. He definitely has everything you want and does look very promising, but I was more impressed by postseason Russel Wilson and RGIII all year than by Luck.

I know he's asked to do more than those 2, he doesn't have a great running game, has a bad OL, but still, I'm not once again convinced he is the next decade top3 QB that everyone sees. I definitely think he can be that, for sure, but I'm not ready to crown him yet. Is anyone feeling the same?

Also, Kevin Kolb between Rodgers and Brady, a sign of things to come in Buffalo?

20
by theslothook :: Tue, 04/09/2013 - 4:19am

As a colts fan - I think he is being overhyped at present. That said, I'm very high on his future value. It really depends on what context you are judging him. Wilson and Rg3 play in offenses that accentuate their qualities while shielding the natural growing pains - ie innovative offenses that rely a ton on strong run game, movement and deception combined with defenses that give them time to operate effectively. Both are stellar and deserving of the praise they are getting.

I will say, the colts were not a good team last year. Their defense sucked, but even their offense wasn't that good. Luck had to carry the team and in doing so, he made a ton of mistakes. His accuracy was inconsistent and he forced a lot of throws. In a sense, he went through a baptism by fire. Why I am high on luck is I think that he needed to go through that. He also displayed some incredible skills that are common to veterans - progression reading, pocket movement, etc. I suspect, much like his predecessor, you need to make a lot of mistakes early if you're going to adapt later on. Most colts fan believe, once the talent improves, luck will only shine further as he will no longer be encumbered but enhanced the way wilson and rg3 are now.

22
by turbohappy :: Tue, 04/09/2013 - 9:56am

I'm not sure what to think. Luck has a bunch of qualities that rookies aren't supposed to have (incredible pocket presence, leadership, etc.), but he too frequently just misses on easier throws with little pressure. He seems like he should have the potential to be great if he is as hard of worker as Manning was, but he's still quite a ways away.

5
by Brent Hutto (not verified) :: Mon, 04/08/2013 - 4:24pm

What is the correlation between a quarterback overall value metric and his adjusted interception rate?

Do good/valuable quarterbacks systematically avoid interceptions or do Brady, Rodgers and the Denver Manning just happen to fall near the very bottom of that list while "QB ARI" and "QB CLE" float to the top...

10
by Perfundle :: Mon, 04/08/2013 - 5:22pm

Oooh, I had been looking for a stat like this! But I was way off in my random guess that dropped interceptions and tipped/desperation interceptions would be roughly equal in number, when actually the former outnumbers the latter by more than 4 to 1.

The next logical step would be to compare the adjusted interception rate to the expected adjusted interception rate based on the opponents played. For instance, P. Manning and Schaub's interception rates are depressed because of the weak pass defenses in the rest of their conference, whereas the entirety of the NFC West's got boosted because all four are top-8 pass defenses.

11
by shah8 (not verified) :: Mon, 04/08/2013 - 6:42pm

Vick has had two tipped interceptions, both off of Celek, one against Baltimore and one against New Orleans. I'm pretty sure neither were catchable if they had not been tipped by Celek, particularly the NO pick 6. And this was just stuff I could remember off-hand! How clean is that dataset, anyways?

18
by chuckdeeznuts (not verified) :: Tue, 04/09/2013 - 2:13am

I don't understand how Wilson was only credited with 2 tipped INTs. There were at least 2 in just SEA@STL.

25
by Aaron Schatz :: Tue, 04/09/2013 - 10:45am

I am happy to look at any play and review it. I would need the game, quarter, and time of play.

We have Wilson with one tipped INT in Week 4 vs. STL, on a pass to Doug Baldwin in the second quarter.

We don't have either of those passes to Celek listed as dropped interceptions. On the Baltimore pass, there was no drop. The pass was defensed when Ray Lewis made contact with the receiver, which caused the ball to be tipped in the air. To be removed from adjusted interceptions, the pass has to be screwed up by the receiver only, without a play being made by a defender. For New Orleans, we have the ball listed as overthrown. I don't have time to look right now at this specific play, but we're not going to count it as a "tipped interception" if you have a bad pass that glances off the fingertips of a jumping receiver. It has to be a good pass that the receiver messes up.

29
by shah8 (not verified) :: Tue, 04/09/2013 - 7:36pm

Yes, Celek was held @ NO, so a kind of pass defended sort of incident as well. The pass, however, was where it was supposed to be.

But the explanation just seems quite dubious as far as telling how good a QB is from getting intercepted vs random bad luck. Most QBs don't have the option that Alex Smith had such that he could prefer to throw to relatively open people. When Alex Smith *did* have to make plays due to the game situation, I remember from the Vikings game just how poorly he played. Different systems make different demands on QBs. Some go for lots of timing throws. Some go for outmuscling (big, fast recievers, big-armed QB throwing darts, such as Cutler to Marshall). Some go for entirely play action. The risk-reward profile is pretty different. Which is why I can't help but think that AY/A is pretty much the only stats that means anything wrt to QB'ing. Anything much more specific can't really say very much.

30
by shah8 (not verified) :: Tue, 04/09/2013 - 7:51pm

I should add, that the pass to Celek @Baltimore was one of those stick throws big time QBs with the rocket arm has to be unafraid to throw. It's a big reason WHY you want QBs with howitzers for arms. Defenses will usually recognize the situation, and coverage will usually be tight. These things usually go to big TEs, and their job is to make catches within tight spaces. Bleeding Green Nation (Eliot Shorr Parks) also saw this play as more Celek's fault.

http://www.bleedinggreennation.com/2013/2/11/3974912/breaking-down-micha...

34
by NeShep (not verified) :: Sun, 04/14/2013 - 11:58pm

Off the top of my head one INT was in and out of Balwin's hands in the first Rams game, another was thrown when his TE fell down without contact while the ball was in the air during the same game, and another in and out of the hands of Lynch in the Dolphins game.