Writers of Pro Football Prospectus 2008

21 Jul 2008

FO Mailbag: Red Zone Interceptions

Zach Z.: Not that it's a great measure, but do you have any idea where I could find a statistic that measures how many Interceptions each quarterback has thrown in the red zone? I have a co-worker who is against Favre coming back to Green Bay, and after I debunked most of his reasons he's hanging his hat on "Favre throws too many interceptions in the red zone," despite having no idea if he's actually higher or lower than any other quarterback.

There doesn't seem to be anything special about Favre throwing more interceptions in the red zone than he does elsewhere. He had only one red zone interception during the 2007 regular season. He had four the year before, which was second in the league, but he had only two in 2005 (the year he led the league in total interceptions by leaps and bounds) and only two in 2004.

In fact, there's no evidence that any quarterback has a particularly unique record of throwing interceptions in the red zone (as opposed to elsewhere on the field). Eli Manning tied for the league lead with four red zone interceptions last year. He didn't throw a single one in 2006. No quarterback has thrown more than two red zone interceptions in each of the last two years, and only four quarterbacks threw more than one red zone interception in both seasons: Jon Kitna, Tony Romo, Ben Roethlisberger, and Peyton Manning. Not exactly a Murderer's Row of Suck, is it? In fact, Peyton Manning has nine red zone interceptions between 2004 and 2007, tied with Favre and brother Eli for the highest total in the NFL over that four-year span.

If your friend thinks red-zone interceptions are the main reason why Favre shouldn't come back, I hope he doesn't try arguing football with anyone from Indianapolis.

Posted by: Aaron Schatz on 21 Jul 2008

27 comments, Last at 25 Jul 2008, 4:51am by ukRaider

Comments

1
by and some others (not verified) :: Mon, 07/21/2008 - 6:21pm

Useful stats for identifying poor quarterbacks:

-Red zone interceptions
-Hair color
-Shoe size
-Tendency to give high fives instead of butt pats (or vice-versa!)

2
by Raiderjoe (not verified) :: Mon, 07/21/2008 - 7:21pm

Fare does not throw any more Ints in erd zone than any other Qb, Where does that guys coworkers come up with such a idea.?
Favr going to be good in 2008 whether with Greeb Bay Pakcers or Washigton Redskins. I think Redskins is good place for Fave to play espeiclaly if team can work out good tarde that doesn't involve Redeskins having to tarde a valuable player. team has good young receivers (M Sweed and D Thomas) and old one (Santana Moss) and some other guys too that aren't bad. Also signed Fred Davis the other day,. Good pickup but I dont know him too well yet.

3
by Bobman (not verified) :: Mon, 07/21/2008 - 7:22pm

That tendency for Peyton to do this is one of the reasons I used to cringe when they'd get inside the 10. (I suspect it's improved post-Edge. See Addai's clinching TD in the AFCCG in 2007--a run not a pass.) Not a big deal, though, since it only pops up in big games: against the Chargers or Pats. I suspect those two teams account for half of them. Jags, too?

My mental image of a typical botched first drive of a game for the Colts is Manning smoothly driving them 75 yards down the field and you start to think "holy cow, they're gonna score a hundred!" and then getting picked in the EZ. They come back to score 30 afterwards, but that first one is just a kick in the nuts. Not sure why it happens. 5 years ago I'd claim "he's trying to do too much," but since he brought his INT count down to about 10 a few years ago, clearly he's been more careful. It's a mystery. And still, I cringe.

4
by Brian (not verified) :: Mon, 07/21/2008 - 7:50pm

RZ ints are very random to begin with, as pointed out. The sample sizes are so small, you'd need very many years of data to tell a QB good at avoiding RZ ints from a bad one.

Plus, there is an inherent bias in the analysis. Better QBs will put their team in the RZ more often, and therefore have more opportunities for an int. You'd need to look at RZ *int rate*, not total RZ ints.

5
by Temo (not verified) :: Mon, 07/21/2008 - 8:40pm

Bobman: Per Brian's explanation- I don't think Peyton Manning has a tenency to throw INTs in the red zone, it just that he's there more often than most QBs. And probably passes more than most QBs in the red zone (though I can't back that up with stats), and so has more opportunities to get picked.

6
by Dan (not verified) :: Mon, 07/21/2008 - 9:02pm

I found some data on red zone offense (linked under my name). The Packers offense was 12th in points per red zone possession in 2007, 31st in 2006, 24th in 2005, and 9th in 2004. Over the same 4 years (2007-2004), the Colts have been 8th, 2nd, 9th, and 5th. That helps vindicate Manning, and (to verify Temo's point), the Colts have also been one of the top 2 teams in number of red zone possessions for each of the past 4 years, with 60+ possessions each year.

To cover the other players that Aaron mentioned, the Giants have been 17th (2007), 10th (2006), 22nd (2005), and 28th (2004) in red zone offense. In the past 2 years, Detroit was 23rd and 27th, Dallas was 14th and 8th, and Pittsburgh was 4th and 24th. There doesn't seem to be much of a relationship between red zone INTs and red zone scoring efficiency, although we'd need to look at more data to be sure.

7
by Temo (not verified) :: Mon, 07/21/2008 - 9:49pm

That's a nice set of data there Dan, so I thought I'd take a crack at it real quick:

Best 5 Teams from 2003-2007(5 seasons) in terms of TD% in Red Zone:
1. San Diego (63%)
2. Kansas City (62.40%)
3. Seattle (61.02%)
4. New England (60.07%)
5. Indianapolis (59.5%)

Worst 5 Teams from 2003-2007(5 seasons) in terms of TD% in Red Zone:
1. Buffalo (63%)
2. Baltimore (62.40%)
3. Cleveland (61.02%)
4. Chicago (60.07%)
5. San Fransisco (59.5%)

Best 5 Teams from 2003-2007(5 seasons) in terms of Score% (TDs AND FGs) in Red Zone:
1. New England(89.77%)
2. Seattle (89.37%)
3. Indianapolis (88.79%)
4. Cincinnati (88.60%)
5. San Diego(88.28%)

Worst 5 Teams from 2003-2007(5 seasons) in terms of Score%(TDs AND FGs) in Red Zone:
1. Miami(77.52%)
2. New York (78.57%)
3. Oakland (79.89%)
4. Chicago (80.91%)
5. Minnesota(81.91%)

8
by Temo (not verified) :: Mon, 07/21/2008 - 9:52pm

Oops, obviously the %'s for the second set of 5 are wrong, they should be thus:

1. Buffalo (41.78%)
2. Baltimore (42.79%)
3. Cleveland (45.37%)
4. Chicago (45.45%)
5. San Fransisco (45.67%)

9
by Temo (not verified) :: Mon, 07/21/2008 - 10:02pm

Here's something else that is interesting. From what I can gather, I would say that simply scoring in the red zone (ie not turning the ball over, not missing a chip shot FG) is NOT correlated from year to year, whereas scoring TDs in the red zone IS.

The correlation from year to year for score % (TDs and FGs divided by possessions) goes: 0.24, 0.18, 0.15, -0.09 from 2003/4 till 2006/7. Also, the league % over that time is 84.4%, with a standard deviation among the teams of 4.17%. In other words, it does not vary very much from team to team, and this year's score % isn't well correlated with next year's.

On the other hand the correlation coefficients from year to year of TD Pct (TDs divided by Possessions) is 51.9% with a standard deviation among the teams of 9.14% (almost double the deviation of simply score %). Also, the correlations from year to year goes: 0.51, 0.367, 0.45, 0.23 from 2003/4 till 2006/7.

By the way, the closer the correlation coefficient is to 1, the greater the correlation between two sets of data. Here I used one year versus the next year for the league by team. I'm not entirely sure if that's what Aaron does, but it seems logical enough to me.

10
by Temo (not verified) :: Mon, 07/21/2008 - 10:53pm

For those are still interested-- it looks like success in scoring Red Zone TDs is well correlated to DVOA (yea, I know... duh!).

From more correlated to least correlated to Red Zone TD%: Total VOA (0.52), Total DVOA (0.51), Pass VOA (0.47), Pass DVOA (0.46), Rush VOA (0.43), Rush DVOA (0.42).(These are the average correlation for the 5 years)

As you can see, they are all well correlated to Red Zone success, though Pass is more correlated than Rush, and Voa moreso than DVOA. The latter makes sense, since when you're talking about an absolute scoring system (whether or not you score a TD), you're more concerned with the non-defense-adjusted metric.

The former is somewhat more interesting: conventional wisdom (perhaps) says that running the ball is more important inside the red zone than passing the ball. This maybe debunks that myth, although the correlations are close enough that I don't think a team that can pass well and run shitty will do markedly better than a team than can run well and pass shitty when inside the red zone.

11
by BOOM!ImChrishHanshen (not verified) :: Tue, 07/22/2008 - 12:11am

Temo,

Nice work in #7. The heading of charts 1 and 3 that you made could also be titled "Teams With Highly stable Quarterback Situations," and vice-versa for charts 2 and 4.

And some others.

12
by Rocky the Philly Eagle (not verified) :: Tue, 07/22/2008 - 12:16am

The Real Question is how does the Madden Cover Jinx apply to Farve?

Did his (un)retirement satisfy the football gods?

13
by Raiderjoe (not verified) :: Tue, 07/22/2008 - 1:47am

re: 18
Maybe Maden game not too late for cover change. game looks stupid if favr gets traded to redskins which is good pososibitlity. Redskins could trade T Collins there and sent j Campbell to bench for one year for more seasoning. fare gets job for one year while Campbell bake in oven. Reskins make run at Superbowl. Maybe get rematch of SB 18 when Raiders kick Redksins asses 38-9. Or maybe get Redskins vs Colts which is new one.
Then Far e r etires and Campbell gets old job back. Collins backups Rodgers in Gb for 2008, both leave in 2009 and Brian Broom Qb Lousiville take over job in 2009. Matt Flymm backup top Broom.
Everything works out good for Gb ans Was under the Raiderjoe plan. Maybe someone should make call tol GB and WAs front offices. Maybe I will step up and do that tomoroow if have time.
Word. you better believe I'm Phat.

14
by Joe T. (not verified) :: Tue, 07/22/2008 - 9:19am

#13 "Everything works out good for Gb ans Was under the Raiderjoe plan. Maybe someone should make call tol GB and WAs front offices. Maybe I will step up and do that tomoroow if have time.Word. you better believe I’m Phat."

That's just gold. Raiderjoe, you are a gentlemen and a scholar.

15
by Aaron (not verified) :: Tue, 07/22/2008 - 9:33am

I'd be really surprised if Washington makes a move for Favre after already trading away some picks to get Jason Taylor.

Trading the future for a couple years out of a great player is one thing, doing it twice in the same season smacks of the kind of thinking that crippled the Redskins for years and that they seemingly have moved away from.

16
by Independent George (not verified) :: Tue, 07/22/2008 - 11:01am

Are red zone interceptions really that much worse than interceptions outside the red zone? While it usually means taking points off the board, but it also means giving up far less field position. Not that I advocate throwing the ball to the wrong team, but if I had to choose, I'd rather be intercepted on the opponent's 10 yard line than my own.

17
by Temo (not verified) :: Tue, 07/22/2008 - 11:56am

Assume having the ball at the 10 yard lines gives you a value of X points, and that having at your own 10 yard line gives you a value of Y points. Turning the ball over on your own 10 yard line costs you Y points while giving your opponent X points. That's a score change of X+Y. Turning the ball over on your opponent's 10 yard line costs you X and gives your opponent Y... same thing, lost value is X+Y.

This is, of course, assuming that the turnover results in field position at the 10 yard line. In practice, I'd assume that opponents are much more likely to score on a return when you turn over the ball on your own 10 yard line, which changes the value gained by your opponent. But I've never seen actual stats for that.

18
by Nathan (not verified) :: Tue, 07/22/2008 - 12:25pm

But by throwing a red zone interception, you lose what should be a gimmie 3 points. So you're not just losing field position, but losing 3 points and the potential for 4 more.

19
by Temo (not verified) :: Tue, 07/22/2008 - 1:01pm

18. And by turning the ball over deep in your own zone, you're giving those same points to your opponenent. The perfect field position valuation metric would encapsulate the total expected points of having the ball at a certain yard line on a certain down/distance.

20
by The Hypno-Toad (not verified) :: Tue, 07/22/2008 - 1:23pm

Wow. I just love this site. Perfect blend of super-genius level football analysis and the feel of a bunch of people hanging out and having fun. A rather innocuous question about someone's mistaken impression about their team's qb led to all of that awesome analysis in the comment thread leading to new questions and leading to new awesome analysis. This is something I've seen over and over on this site, but every once in a while, I feel the need to comment on it, because it just makes me feel all warm in my football-heart.

21
by Wile E. Coyote (not verified) :: Tue, 07/22/2008 - 2:28pm

Did someone call me?

22
by ukRaider (not verified) :: Wed, 07/23/2008 - 12:54pm

The way I see it, the only potential use for such a stat (i.e. Red Zone Interception Rate) is to identify a QB who is especially skilled at throwing short/medium distance passes, in a congested area, and which are out of (easy) reach from defenders (I'll refer to this as Exceptional Red Zone Skill, or ERZS). Assuming a reasonably large enough baseline sample size (whatever that may be), I don't think you will ever find a QB who has an especially high rate of INTs in the RZ; a QB who clearly has such a severe weakness wouldn't have been allowed to play enough games to have a large enough sample size.
However, if you were to find a QB who has an especially low rate of INTs in the RZ, that would be the clearest possible indication of ERZS. For most QBs (possibly all of them), I don't think the differences in their rates are meaningful enough to say, for example, that QBa is 22% more skilled in the RZ than QBb. This is due to the fact that pass attempts in the RZ are infrequent plays (relative to all other plays). So although the rates can be accurately measured, I don't think the implications of them would be nearly as accurate in terms of measuring relative RZS between QBs.

23
by Temo (not verified) :: Wed, 07/23/2008 - 4:08pm

22. Your analysis is supported by some of the data I posted above. I found that score% (or, to put it another way, given the rarity of a missed FG in the Red Zone... % of time that a team doesn't turn over the ball or fail on 4th down/run out of time) isn't very correlated from one year to the next and the difference between the best teams and the worst teams isn't very great. TD% however is much more variable between teams and the correlations are much higher.

24
by BillWallace (not verified) :: Wed, 07/23/2008 - 6:04pm

I too hope the Redskins don't tard a valuable player.

25
by ukRaider (not verified) :: Thu, 07/24/2008 - 4:40am

23. I'm not sure that year-to-year statistical correlations in general offer much information about a team. Teams don't often stay the same as they were the year before (i.e. without significant changes), so any year-to-year correlations may only be an indication how well or poorly a team is being run from the top down, how sound a particular part of the team is being managed, or may just be coincidental.

What the RZ Score% chart’s tells us should surprise no-one, which is that even the worst offensives will come away from the RZ with at least a FG 4/5 times.

RZ TD% shouldn’t offer anything noteworthy either; it will just be a reflection of the overall offensive efficiency, which can be more precisely measured with various other stats. You may find a consistently good performing offense (say, over the course of at least one season) that is consistently poor in RZ TD%, which would indicate they have particular trouble running or passing into a congested area. But I doubt you would find a consistently great offense that is consistently poor in RZ TD%.

Aside from identifying a QB with ERZS as I mentioned in my previous post, I don't see any RZ measurement that can tell us something that cannot be better illustrated using other offensive stats.

20. Nice post btw. I think I speak for the majority when I say "Ditto".

26
by Temo (not verified) :: Thu, 07/24/2008 - 12:08pm

Year-to-year correlations are how FO gets a lot of its predictive stats from, although the concerns you raise are well founded. However, these are concerns for ANY correlatively founded statistic. Correlation doesn't show cause, as a rule.

RZ TD% IS correlated VERY well with VOA, so you are right in that way. However, the interesting point that i made was that it was (very little, but still there) more correlated with passing VOA than Rush VOA, though total VOA is still most correlated.

27
by ukRaider (not verified) :: Fri, 07/25/2008 - 4:51am

I was talking about looking for correlations between two different stats in back-to-back years (that was what came to mind when you mentioned "one year to the next") in order to learn/predict something about a specific team. So for example, if you were to find a strong correlation between any stat in Year 1 and wins in Year 2, I can't see how you could extract a meaningful conclusion from that. However, having re-read your post (23) I now realise that I am not certain exactly what you mean by "Score% isn’t very correlated from one year to the next"; what other variable are you correlating it with?

RE: Passing VOA being slightly more correlated with RZ TD% than Rushing VOA; My initial thoughts on this are that teams in general tend to rely slightly more on the passing game for TD attempts in the RZ, and that might explain the slightly stronger Passing VOA correlation you found. I'm relatively new to the game so my perception on actual trends isn't strong, but I'd be curious to see the RZ Pass vs RZ Rush attempt totals from last year.