Here are the Week Seven Quarterback and Running Back Rankings:

**Table One: Quarterback Rankings for Week Seven**

**Table Two: Running Back Rankings for Week Seven**

**Checking in on the Favre Trio**

Perhaps this is as good a time as any to review the recent game of musical chairs – at the quarterback position — played by Green Bay, the New York Jets, and the Miami Dolphins. After seven weeks, here are the numbers on the quarterbacks involved:

- Chad Pennington: 0.316 net points per play (ranked #2)
- Aaron Rodgers: 0.235 net points per play (ranked #12)
- Brett Favre: 0.100 net points per play (ranked #24)

An average quarterback posts a per play mark of 0.159. So Pennington and Rodgers have been above average this year while Favre has not. Much of the difference in these players can be linked to yards (passing yards + rushing yards – yards lost from sacks) per play (passing attempts + rushing attempts + sacks). Here is where these quarterbacks rank in this stat:

- Chad Pennington: 7.21 yards per play (ranked #4)
- Aaron Rodgers: 6.45 yards per play (ranked #11)
- Brett Favre: 5.53 yards per play (ranked #22)

If we turn to the other primary element of The Wages of Wins quarterback metrics – turnovers (interceptions + fumbles lost) per play – we see that Pennington continues to lead this trio:

- Chad Pennington: 1.579 turnovers per 100 plays (ranked #2)
- Aaron Rodgers: 2.273 turnovers per 100 plays (ranked #10)
- Brett Favre: 4.545 turnovers per 100 plays (ranked #31)

In sum, whether we look at efficiency in moving the ball (yards per play) or the propensity to give the ball away, Favre is lagging behind the player he replaced in New York and the player who replaced the legend in Green Bay. Of course, quarterbacks are very inconsistent. So this ranking can certainly change before the season is over.

**Predicting One Turnover**

Speaking of turnovers, Jason Campbell is the only quarterback who has not lost the ball this season. And this week he plays the Detroit Lions. Detroit and Seattle are the only teams who have only intercepted one pass this year. Given these two stats, one would expect Campbell to keep his interception record clean for one more week.

One should note, though, that turnovers – as I noted in The New York Times about two years ago – are not predictable. How much a quarterback turns the ball over per play last season tells us very little about his propensity to give the ball away this season. This tells us that what Campbell and the Lions have done thus far this season is not going to give us much insight into the game on Sunday.

Consequently, I am going to make a bold prediction. I am going to watch the entire Detroit -Washington game (much to my wife’s chagrin). And I expect to see Campbell throw at least one interception. This is technically a guess (and a hope). But Campbell is probably going to throw an interception at some point this season and I see no reason why it can’t happen against the Lions.

Two more notes on the Lions… Dan Orlovsky and Kevin Smith – Detroit’s starting quarterback and running back – do not have enough attempts this year to be ranked. But if they were ranked, each wouldn’t look too bad. Orlovsky’s net points per play stands at 0.152, a mark that would rank 21^{st} (out of 34) quarterbacks in the league. This is not a great mark by NFL standards, but it’s pretty good for a Detroit quarterback.

As for Smith, the Lions are known for great running backs (see Barry Sanders and Billy Sims). Smith may not become Sanders or Sims, but he’s off to a good start. Per play, Smith is producing 0.204 points, a mark that would rank fourth in the NFL. And Smith has done this behind nine different starting offensive linemen.

So although the Lions are 0-6, there are some bright spots. Detroit has Kevin Smith, Calvin Johnson, and….. Okay, they have Smith and Johnson. And a defense that I really think will get one interception this week.

– DJ

The WoW Journal Comments Policy

For more on the Wages of Wins football metrics see

**Consistent Inconsistency in Football**

**Football Outsiders and QB Score**

**The Value of Player Statistics in the NFL**

*Basketball Stories*

Brian

October 25, 2008

Inter-season turnovers are not consistent, but intra-season offensive interception rates are somewhat consistent.

If you divide the NFL season in half: games 1-8 and 9-16, team interceptions per attempt auto-correlates at 0.27. It’s not a lot, but it’s something, and it’s a lot more than the season-to-season correlation which is near zero.

Also, if teams that had QB changes were excluded, the intra-season correlation would likely be a little stronger.

So while Campbell will almost certainly throw a few ints this season, chances are good he’ll continue to have a below-avg int rate for the remainder of the season.

dberri

October 25, 2008

Brian,

0.27 is still really low. This means that only 7% of interceptions in the second half are explained by what happened in the first half. I think it was you who note that the Lions 6-2 start last year was a mirage since it was based on a very low rate of turnovers (which could not be counted on to continue). So I am still hoping Detroit will get an interception tomorrow.

Brian

October 26, 2008

Agreed. It’s not high, but considering the rarity of the event, a couple interceptions either way can add a lot of variance.

Ints “thrown” and “taken” are different. Def ints auto-correlate even lower. 0.08 if I recall correctly. To me, this suggests that yes, ints are very random but also that they are “thrown” by an offense far more than they are “taken” by a defense.

Good luck to your Lions!

Brian

October 26, 2008

Dave-I’m a little confused, why do you say 7% of the variable is explained by an r of .27? Doesn’t a correlation of .27 mean 27% of the second variable is explained by the first? Correlation is r, not r^2.

I think of r as “for every standard deviation above or below the mean in the first variable, we could expect r standard deviations above or below the mean for the second variable.”

So if offensive int rate has a mean of 3.1% and a SD of 0.8%, a team that is currently 1 SD below the mean in int rate (2.3%) could be expected to have:

3.1 – ( .27 * .8) = 3.1 – .2 = 2.9%

for the 2nd half of the season. In other words, 27% of the deviation from the mean carries across the 2 halves of the season, not 7%.

dberri

October 26, 2008

R-squared is the percentage of the dependent variable that is explained by your model. In this case your model is Second half interceptions = f(first half interceptions). If the r = 0.27, then the r-squared = 0.27 * 0.27 = 0.07. In sum, 7% of second half interceptions are explained by the first half interceptions.

By the way, as I watch the Redskins march down the field, I am not sure the Lions will ever get an interception (or force a fumble, punt, etc…). The Lions are just bad.

Brian

October 26, 2008

Ok. I see what you’re doing now. But r^2 isn’t the percentage of the dependent variable explained by a model, it’s a percentage of the “variance” of a dependent variable explained by a model.

Although the variance would be .07, the units of variance would be “interceptions per attempt

squared.” That is only an abstract statistical unit and has no concrete meaning.Sure, 93% of the variance is explained by other things, but those other things include covariance with interception rate itself. So if we’re trying to predict real 2nd half season interceptions per attempt, we’d use r, not r^2. In other words, you’re shouldn’t be worried about the Lions’ interception variance as much as their SD from the mean.

I guess it’s just a matter of how you look at it, but I’ll defer to your interpretation, especially because I feel sorry for your Lions after Santana Moss’s punt return!

dberri

October 26, 2008

Brian,

Yes, we are explaining variation.

I have no idea what you are saying when you say “interceptions per attempt squared”. How I define r-squared is the standard definition in every econometric textbook. R-squared tells us the percentage of the variation in the dependent variable (not the dependent variable squared) that is explained by the model.

Looks like I am not going to be wrong about the Lions getting an interception. They did force a Campbell fumble, and this is the first turnover by Campbell this season. So that’s something (although I think we are still going to lose).