Writers of Pro Football Prospectus 2008

Most Recent FO Features

BrownMal15.jpg

» Futures: Texas RB Malcolm Brown

DeMarco Murray is the toast of the NFL, but injury and team issues clouded some observers' view of his talent. Texas RB Malcolm Brown might have the same problem this winter. 

07 Aug 2014

Adjusted Interceptions 2013

by Aaron Schatz

Well, we're kind of late on rolling out some of our 2013 stats, but better late than never, right? Let's take on the wonderful world of adjusted interceptions.

Here's the basic idea, which we first introduced three years ago:

  • 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). Starting in 2012, we also included plays where the defender didn't straight-out drop the ball but instead had it knocked out of his hands by an offensive receiver (a "defensed interception").
  • We subtract plays where the interception (or a dropped 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.

Interceptions are notoriously hard to forecast from year to year, because there's so much random chance and statistical noise involved. Historically, we've found that adjusted interceptions have a higher year-to-year correlation than standard interception totals and are a better predictor of future interceptions. For the entire period from 2007 through 2013, the correlation coefficient for adjusted interception rate (0.32) is twice as high as the correlation for actual interception rate (0.16). In set of years except one (2011-2012), adjusted interception rate was a better predictor of interceptions the next year than actual interception rate.

In 2013, for the most part, the quarterbacks with a lot of dropped interceptions were the same as the ones with a lot of actual interceptions. Ryan Tannehill, who had 17 actual interceptions, led the league with 11 dropped interceptions. He had the league's biggest gap between actual interception rate (2.9 percent) and adjusted interception rate (4.5 percent).  Geno Smith ended up with the highest adjusted interception rate, 5.9 percent, throwing 21 picks with five dropped interceptions but no "tipped interceptions" or Hail Mary interceptions. Eli Manning had only five dropped interceptions to go with his 27 actual interceptions, but with no "tipped interceptions" and only one Hail Mary, he still ended up with the second highest adjusted interception rate at 5.6 percent.

In the other direction, the biggest difference between interception rate and adjusted interception rate belonged to Tom Brady. Brady threw 11 interceptions last season, but had only one dropped interception. He also had two picks that were tipped by receivers and two interceptions which were effectively no different from incomplete passes: the famous play where Luke Kuechly interfered/did not interfere with Rob Gronkowski in the end zone, and an interception thrown to the end zone on fourth down with seven seconds left in a Week 15 loss to Miami. That all works out to just eight adjusted interceptions, a miniscule 1.3 percent adjusted interception rate.

Two other players had just a 1.3 percent adjusted interception rate, but they're the players who already had a very tiny actual interception rate with small sample sizes: Josh McCown (one pick plus two dropped picks) and Nick Foles (two picks plus two dropped picks).

Another player with a big difference was Mr. I Played The Hardest Defensive Schedule, Matt Ryan. The big thing for Ryan was that he had "interception as good as an incomplete" plays: Week 1 when he threw a pick in the end zone on fourth-and-goal with 49 seconds left; the Hail Mary in Week 16's loss to San Francisco; and one that doesn't normally count by our "last 2:00" rules but I counted anyway, when the Falcons were losing by 14 points with 3:08 left in Week 8 and Ryan threw a pick on a desperate fourth-and-19.

The following table lists adjusted interceptions for all quarterbacks with at least 200 pass attempts. Note that most quarterbacks will have more adjusted interceptions than actual interceptions. The leaguewide interception rate is 2.70 percent and the leaguewide adjusted interception rate is 3.26 percent. Therefore the blue "difference" column represents the difference between the two rates compared to that -0.56 percent average, so that positive always means "threw fewer picks than expected" and negative means "threw more picks than expected."

Player Team INT HM/
End Q4
Drop/
Def INT*
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate Difference
(vs. Avg)
E.Manning NYG 27 1 5 0 31 551 4.9% 5.6% 0.2%
J.Flacco BAL 22 0 9 2 29 613 3.6% 4.7% 0.6%
M.Stafford DET 19 0 8 0 27 631 3.0% 4.3% 0.7%
R.Tannehill MIA 17 2 11 0 26 584 2.9% 4.5% 1.0%
C.Palmer ARI 22 0 4 0 26 573 3.8% 4.5% 0.1%
G.Smith NYJ 21 0 5 0 26 442 4.8% 5.9% 0.6%
A.Dalton CIN 20 0 3 2 21 585 3.4% 3.6% -0.4%
C.Henne JAC 14 0 7 0 21 502 2.8% 4.2% 0.8%
M.Schaub HOU 14 0 5 0 19 358 3.9% 5.3% 0.8%
C.Newton CAR 13 0 5 0 18 470 2.8% 3.8% 0.5%
M.Ryan ATL 17 3** 2 0 16 650 2.6% 2.5% -0.7%
B.Roethlisberger PIT 14 0 2 1 15 579 2.4% 2.6% -0.4%
R.Griffin WAS 12 0 4 1 15 456 2.6% 3.3% 0.1%
Player Team INT HM/
End Q4
Drop/
Def INT*
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate Difference
(vs. Avg)
D.Brees NO 12 0 3 1 14 650 1.8% 2.2% -0.2%
A.Luck IND 9 0 5 0 14 574 1.6% 2.4% 0.3%
P.Rivers SD 11 0 2 0 13 543 2.0% 2.4% -0.2%
J.Cutler CHI 12 1 2 0 13 352 3.4% 3.7% -0.3%
R.Fitzpatrick TEN 12 0 2 1 13 347 3.5% 3.7% -0.3%
P.Manning DEN 10 0 2 0 12 658 1.5% 1.8% -0.3%
A.Smith KC 7 0 4 0 11 504 1.4% 2.2% 0.2%
M.Glennon TB 9 1 3 0 11 417 2.2% 2.6% -0.1%
T.Pryor OAK 11 0 1 1 11 271 4.1% 4.1% -0.6%
B.Weeden CLE 9 0 2 0 11 266 3.4% 4.1% 0.2%
C.Ponder MIN 9 1 4 1 11 240 3.8% 4.6% 0.3%
T.Romo DAL 10 1 2 1 10 535 1.9% 1.9% -0.6%
C.Kaepernick SF 8 1 3 0 10 416 1.9% 2.4% -0.1%
Player Team INT HM/
End Q4
Drop/
Def INT*
Tip INT Adj INT Pass Att
(no DPI)
INT Rate Adj Rate Difference
(vs. Avg)
E.Manuel BUF 10 0 3 2 10 305 3.0% 3.3% -0.2%
M.Cassel MIN 9 0 2 1 10 254 3.5% 3.9% -0.2%
M.McGloin OAK 8 0 2 0 10 211 3.8% 4.7% 0.4%
R.Wilson SEA 9 2 3 1 9 407 2.2% 2.2% -0.6%
K.Clemens STL 7 0 3 1 9 241 2.9% 3.7% 0.3%
T.Brady NE 11 2 1 2 8 624 1.8% 1.3% -1.0%
J.Campbell CLE 8 0 1 1 8 316 2.5% 2.5% -0.6%
C.Keenum HOU 6 0 3 1 8 251 2.4% 3.2% 0.2%
M.Flynn 2TM 5 0 2 0 7 200 2.5% 3.5% 0.4%
A.Rodgers GB 6 0 1 2 5 289 2.1% 1.7% -0.9%
S.Bradford STL 4 0 2 1 5 262 1.5% 1.9% -0.2%
N.Foles PHI 2 0 2 0 4 317 0.6% 1.3% 0.1%
J.McCown CHI 1 0 2 0 3 225 0.4% 1.3% 0.3%
*Does not include passes which are first tipped into the air by receivers and then turn into dropped/defensed interceptions.
**Includes a deep interception thrown on fourth down with 3:08 left and Atlanta down two touchdowns.

Click here for a look at the adjusted interception numbers from 2012.

Posted by: Aaron Schatz on 07 Aug 2014

19 comments, Last at 27 Sep 2014, 8:08pm by dfobare

Comments

1
by theslothook :: Thu, 08/07/2014 - 5:32pm

"For the entire period from 2007 through 2013, the correlation coefficient for adjusted interception rate (0.32) is twice as high as the correlation for adjusted interception rate (0.16)"

Think there is a typo there somewhere.

2
by Led :: Thu, 08/07/2014 - 5:57pm

"[Brady] also had two picks that were tipped by receivers and two interceptions which were effectively no different from incomplete passes: the famous play where Luke Kuechly interfered/did not interfere with Rob Gronkowski in the end zone, and an interception thrown to the end zone on fourth down with seven seconds left in a Week 15 loss to Miami."

If the ball had not been intercepted on the Keuchly play, it would have been called interference. That's a big difference from an incomplete pass.

5
by DEW :: Thu, 08/07/2014 - 7:18pm

No, if the defender had dropped the ball (making it a 4th down incomplete), it still wouldn't have been interference.

Though in a way, I do agree with you. The usual point is that a 4th down interception is the same as a 4th down incomplete, so shouldn't count any worse against the QB, but in this case, the fact that Brady threw the pass square to a defender instead of out of the back of the end zone, say, actually did have a significant difference in the outcome of the play. The weird exception that proves the rule.

3
by MJK :: Thu, 08/07/2014 - 6:25pm

I think you could probably drop the "final 2:00" stipulation from 4th down INTs. On 4th down, an incomplete pass results in a turnover anyway regardless of the time left on the clock, so an INT is at worst equal to an incomplete, and in practice often better because the defense gives up field position.

In fact, unless the defender has a clear and unambiguous run back lane, he should NEVER intercept a 4th down pass, but rather bat it to the ground. So almost any INT thrown on 4th down would have been incomplete if the defender had played optimally.

6
by Aaron Schatz :: Thu, 08/07/2014 - 7:34pm

In general, an interception on a short- or medium-length pass on fourth down has a chance of being returned past even the original line of scrimmage -- sometimes even for six. That's why we still consider a fourth-down interception as a "full interception" except at the end of the game. If a quarterback is losing in the final two minutes, however, obviously it doesn't matter if the fourth-down pick is returned for a touchdown by the defense. The offense was going to lose that game whether that ball incomplete or intercepted, and no matter how long the interception return was.

19
by dfobare :: Sat, 09/27/2014 - 8:08pm

Does FO keep track of this *during* the season?

4
by dbrude@gmail.com :: Thu, 08/07/2014 - 6:48pm

It would be nice to adjust the INT rates by the average depth of throw as well...

http://www.footballperspective.com/more-work-on-popip-and-predicting-int...

7
by jfsh :: Thu, 08/07/2014 - 8:22pm

"Another player with a big difference was Mr. I Played The Hardest Defensive Schedule, Matt Ryan. The big thing for Ryan was that he had "interception as good as an incomplete" plays: . . . the Hail Mary in Week 16's loss to San Francisco"

This one actually swung the title of our fantasy league. The winner had the SF defense and, after going back and forth all night, was down by 1.5 points with essentially no time left on the clock and nobody left to play on either side. That otherwise-meaningless interception was worth 2 points and gave him the victory (and a rather fat purse)!

8
by countertorque :: Thu, 08/07/2014 - 9:15pm

Awesome article. Counting INT's without acknowledging how noisy the data is has always bothered me.

9
by theslothook :: Thu, 08/07/2014 - 9:39pm

THat's true for just about every stat that is currently tabulated. Dropped Tds, sack's wiped out by penalties/near sacks etc.

10
by Will Allen :: Fri, 08/08/2014 - 12:53am

The Ponderous One managed to combine a high adjusted rate with a large dose of timidity, causing him to look like he was apologizing when throwing downfield. How they ever saw top half of the 1st round talent in that guy is just beyond understanding. The only way for a Vikings fan to optimistic going forward is to hope the owner insisted that a qb be taken, and he has since learned to butt out.

I wonder if there is any positive correlation, and if so (as I suspect), how strong it is, between adjusted rate and a qb's defense giving up a lot of 1st half points.

11
by Red :: Fri, 08/08/2014 - 12:59am

You list Peyton Manning with no tipped interceptions, but I think he had one against Washington. On a throw to the left sideline, Demaryius Thomas had the ball in his hands then literally had it ripped away by DeAngelo Hall. I know it technically wasn't "tipped" but it was still receiver error that caused the INT.

12
by ChicagoRaider :: Fri, 08/08/2014 - 9:09am

Actually, it would be nice to see the table sorted by rate rather than INTs, because throws and playing time are not equal across the players, and the percentages are nowhere close to the number of interceptions because of that.

13
by ChicagoRaider :: Fri, 08/08/2014 - 9:10am

Also, "Playe" should have an "r" at the end in the top of the first column.

14
by mehllageman56 :: Fri, 08/08/2014 - 1:01pm

I'm pretty sure Geno Smith had a pass tipped to a defensive back in the first New England game. Then again, perhaps Clyde Gates just handed the ball off to the defensive back.

15
by wiesengrund :: Sat, 08/09/2014 - 2:59am

Wondering how Ryan ends up with 0 tipped INTs, considering the Bowman pick six at Candlestick?

16
by TomC :: Wed, 08/13/2014 - 1:23pm

Nick Foles threw as many picks in the 1st quarter of last week's preseason game as he did all last year.

18
by dfobare :: Sat, 09/27/2014 - 8:05pm

Does FO keep track of this info *during* the season?