Comments on Advanced NFL Stats: Why the Chargers Defense Will Decline in '08

Hindsight is truly 20/20: well-analyzed, well-writ...

2010-02-02T17:21:24.730-05:00

Hindsight is truly 20/20: well-analyzed, well-written.

Anon-First, if you don't buy that def int rates ar...

2008-09-26T10:43:00.000-04:00

Anon-First, if you don't buy that def int rates are effectively random, see this article.

Second, it sounds like you don't understand what regression to the mean is all about. 'Independent variables' does not mean that each outcome is equally likely. The frequency of interception rates, just like almost any random variable, is clustered around the mean--it's a normal/Gaussian distribution. So it's far more likely that the SD defense will have close to an average int rate than another year of a phenomenally high rate. It doesn't take hundreds of seasons! That's hilarious. It's possible they'll repeat, but very unlikely.

How's SD int rate so far this year? I thought so. Good luck on your Stats 101 redo! Let me know if you pass this time.

Sorry... that's the law of large numbers... there'...

2008-09-26T10:14:00.000-04:00

Sorry... that's the law of large numbers... there's no such thing as the law of large averages or the law of averages.

Your idea that a high scoring team will naturally ...

2008-09-26T10:13:00.000-04:00

Your idea that a high scoring team will naturally "regress to the mean" the next season is erroneous. If interceptions are random, as you claim, then each season's interception stats are independent variables and the teams ability to get high interceptions one year will have no impact on the next years interceptions. The law of large averages implies that the team's average season stats will migrate towards center, but that's over hundreds or thousands of seasons, not one season to the next... so your premise is flawed. This is all stats 101.
Anyway I don't buy that interceptions are entirely random anyway, and the picture on the top of this page proves it. Most defensive players wouldn't even attempt to go for that catch. It's just as much skill and mindset as it is luck.

Its seems to me that another factor causing regres...

2008-08-08T11:41:00.000-04:00

Its seems to me that another factor causing regression to the mean would be once a CB establishes himself as a guy who can pick off alot of passes and return them for TDs, as Cromartie did last year, he might get thrown at alot less.

I don't know how much of it was fact or fiction, but I seem to remember that being a big theory about Dion during his prime, that teams just wouldn't throw to his side for fear of the consequences

If that's what's actually happened back then, or this upcoming season with Cromartie, it would seem like a CB could have a pretty big impact over the game with out ever making a play

The rate at which you cause interceptions isn't ch...

2008-07-29T09:40:00.000-04:00

The rate at which you cause interceptions isn't chaotic and random. The Chargers caused roughly the same amount of would be interceptions in 2006 and 2007.

The rate at which you catch those interceptions is however different. Instead of Drayton Florence dropping 7 would be interceptions in 06, Antonio Cromartie caught them.

If you didn't notice the Ravens didn't sustain the rate at which they caused interceptions with a pass rush that fell off the face of the earth.

While I agree with the randomness argument and dat...

2008-07-27T00:13:00.000-04:00

While I agree with the randomness argument and data, check out the Bolts schedule this year. One of the two easiest IMHO based on facing high powered offenses with strong passing attacks. With the Bolts scoring machine humming along, I'd bet they beat the average INT rate this year, and repeat as a top 5 DST.

Total yards allowed is also very offense-dependent...

2008-07-22T13:37:00.000-04:00

Total yards allowed is also very offense-dependent (your own offense). I'd start with yards per rush and net yards per pass. Both correlate from yr-to-yr at about .25 or more. It's not a lot but its a start.

OK, I'll buy the idea that randomness & re...

2008-07-22T07:16:00.000-04:00

OK, I'll buy the idea that randomness & regression-to-mean render interceptions a poor predictor of future defensive success. What defensive stats would you expect to be better predictors? Not fumbles recovered. It seems like points allowed might also be subject to randomness and the variability of the offense, although perhaps to a lesser extent than turnovers. How about total yards allowed? Punts forced? What else?

Since the offense (QB) drives the interception rat...

2008-07-21T14:21:00.000-04:00

Since the offense (QB) drives the interception rate, are the QBs scheduled to face San Diego's defense this season significantly better/worse than the QBs the defense faced last season?

That Cromartie INT pictured at the top was absolut...

2008-07-20T23:05:00.000-04:00

That Cromartie INT pictured at the top was absolutely insane.
Peyton's 6 INT game was a big part of the Chargers high INT rate. One phenomenal play by Cromartie, one hot route pass tipped straight up by the blitzer, one bad route by a 3rd string WR, one pass tipped by the intended receiver right to a defenser, one picked off Hail Mary. That game gives a great example of the luck/randomness of INTs.