Writers of Pro Football Prospectus 2008

12 Aug 2010

FOA 2010 MEDIA: Dallas

My annual talk with ESPNDallas.com's Tim MacMahon about the Cowboys chapter of FOA 2010 is up.

It addresses the Cowboys projection for this year, our hopes for Miles Austin, and whether Leonard Davis and Andre Gurode really struggled last year.

Posted by: Bill Barnwell on 12 Aug 2010

31 comments, Last at 15 Aug 2010, 11:08pm by thendcomes


by jklps :: Thu, 08/12/2010 - 2:42pm

Why do I get so much pleasure out of the feedback on these media interviews about how wrong and crazy FO's predictions are.

The NFL's ability to market in hope is amazing.

by Temo :: Thu, 08/12/2010 - 3:13pm

To be fair, over the span of the past 5 years, those saying the Cowboys predictions have been too pessimistic have been correct.

Edit: That said, I think this year I agree with FO's prediction. The offense is headed for a decline and the defense was never very good anyway.

by tuluse :: Thu, 08/12/2010 - 3:28pm

That doesn't quite excuse comments like this:

How do you PREDICT injuries? How can he PREDICT Desean Jackson having a field day with our secondary when he was shut down 3 times last year with an experienced qb? How can you PREDICT less than 1000 yards receiving for Miles Austin when he is Tony Romos primary target and throws for over 4000 yards? Amateur article with a biased opinion from a ridiculous, unprofessional blog site.

by jklps :: Thu, 08/12/2010 - 3:29pm

It's not just the Cowboy fan responses I enjoy. I enjoy them all. Mostly because they get so angry out of fandom and not understanding what the predictions mean, and obviously you can only be correct to a degree in these predictions anyways.

by Temo :: Thu, 08/12/2010 - 3:35pm

Seems like reasonable questions. The statement at the end is ridiculous though.

by tuluse :: Thu, 08/12/2010 - 3:47pm

"How can you predict injuries?" is not a reasonable question unless you think predictions in general are just impossible, and if that's the case I don't see why you would read an article on predictions and feel the need to comment. The questionable use of capitalization doesn't really help his case.

by jbrown (not verified) :: Thu, 08/12/2010 - 4:36pm

From what I've seen people in general don't handle predictions and probabilities well. They tend to see it as an all or nothing deal, and expected returns are difficult to comprehend.

This is an off-topic example but maybe someone here will get something out of it: I go back to my alma mater a few times a year to talk to students about starting and growing a business (I'm 26 and have been fortunate to build 2 successful companies, so it tends to go better than some 50+ yr old that can't relate with the "tweeters and interwebs"). Nearly every time I run a simple coin flip test where the class votes on whether the next flip is more likely to be A) heads B) tails or c) they are equally likely...and almost every time there will be heated discussions about whether you should follow the trend or go against the trend, and there is maybe one nerdy kid in the back laughing at the idea. The point of the whole exercise is to show how people over-attribute success to something (like a product in my case) when it really depends on random chance, and there is always a big "AHAH" moment.

I bring it up because I've actually modeled a lot of systems I use on things I've learned from this site, and that exercise would probably last about 30 seconds with this audience.

by Temo :: Thu, 08/12/2010 - 7:24pm

Your answer to the question seems to be mistaken, if I'm understanding you correctly.

He's actually asking how you can predict injuries. FO doesn't "predict injuries", they simply assume that everyone will regress to the same rate of injuries. That would be the appropriate answer.

by dbostedo :: Thu, 08/12/2010 - 7:56pm

It may not be the same rate for every team. Do you happen to know (or maybe one of the FO guys can answer) whether or not they adjust expected injury rate based on the particular team's injury history and/or the team health ratings?

by Theo :: Thu, 08/12/2010 - 10:26pm

...or send the book to the EU!
It's not avsailable here.
Ok someone will send it but it still sucks.

by tuluse :: Thu, 08/12/2010 - 7:58pm

Well tone is hard to convey in text, and I read it as "how is it possible to predict injuries?"

by Temo :: Thu, 08/12/2010 - 10:43pm

Isn't that a fair question though? I'm not trying to be difficult, I'm just saying that predicting injury proclivity has to be one of the holy grails of professional sports and I don't know that anyone's figured it out yet.

FO certainly hasn't, they just assume every team will have approximately the same injury rate (and it's a fair way of doing it, barring additional information).

by tuluse :: Thu, 08/12/2010 - 11:39pm

If you don't think it's possible to predict injuries at all, that's a fine opinion to hold, but then why are you reading articles about NFL predictions and feeling the need to comment on them?

by chemical burn :: Fri, 08/13/2010 - 10:48am

But isn't part of the issue with the Cowboys that it's not just that FO is predicting to have injuries (which they then avoid at an above average rate), but that FO is saying their depth at certain positions like QB, CB, S and O-LINE is notably bad so if those injuries did happen, it would hurt the team more than it would hurt, say, the Eagles (who have suffered through significant injuries to all of those positions in the past few years and STILL had an above average winning percentage)?

I think that's what makes FO's injury projection so hard on the Cowboys - it doesn't see depth there at key positions like QB (whereas losing McNabb for games at a time never really hurt the Eagles) and IF they did suffer slightly above average amount of injuries, they would struggle to win. The Cowboys for several years had anomalously above average success (I don't want to call it luck) in avoiding injuries, so FO consistently under-rated them... And then it says "Next year, for real guys, listen up, next year... they're going to regress to the mean with injuries and it's going to bite them."

I personally buy into the idea that Dallas has a great training staff and can sustain their injury rate and that their success at avoiding injuries isn't luck and, therefore, they are likely to be a 10 win team this year. Unless the Giants get their defense figured out, the Eagles go 11-5 again and McNabb/Shanahan get the Redskins act together. I'd imagine that the Cowboys low projection every year comes from the fact that the NFC East is a tough place to play and certain factors that FO sees as unrepeatable luck (like injury, but also 3rd down success) can't be sustained in such a tough division.

by t.d. :: Fri, 08/13/2010 - 12:01pm

well they've reversed their position about 3rd down success, at least on offense.

by chemical burn :: Fri, 08/13/2010 - 12:38pm

In general or in regards to Dallas? Sorry, I haven't gotten the new FOA because I was out of the country for a month.

by t.d. :: Fri, 08/13/2010 - 12:52pm

In general. In the last five years, they say several teams (Indy, New England, Dallas, I think there were others) have consistently been better on third down than on 1st and 2nd downs. The 3rd down improvement trend still exists for defenses, however.

by chemical burn :: Fri, 08/13/2010 - 2:03pm

Interesting. In that context, disproportionate 3rd down success would seem to be an indicator of team that can "turn it on" so to speak and one likely to succeed in general...

by t.d. :: Fri, 08/13/2010 - 2:43pm

Actual list of teams consistently better on 3rd down was Indy, San Diego, and Green Bay. Teams consistently worse on 3rd down were Jets and Seahawks. I know in the past I've read that Dallas was unsustainably successful on 3rd downs, but that was probably in the 2007 season.

by alexbond :: Fri, 08/13/2010 - 5:55pm

I can almost assure you that the Seahawks will improve on 3rd down since Holmgren left since they will cool it on the 3rd down draw play which almost never works. Holmgren's love for the 3rd down draw, particular the fullback draw, was always his biggest weakness as a playcaller.

by tuluse :: Fri, 08/13/2010 - 2:57pm

I would say from observation this makes sense with the Colts. They run plays they know won't be very successful on first 1st and 2nd down. No running back is going to be as good at moving the ball as Manning is. Then on 3rd down they use their money plays.

by t.d. :: Fri, 08/13/2010 - 1:37am

FO preseason projections have not predicted the Cowboys to have a winning record any of the last five seasons, and the Cowboys have had a winning record in all of those seasons.

by Richie3Jack (not verified) :: Fri, 08/13/2010 - 9:20am

I honestly think that FO.com hasn't really scratched the surface on using statistics to project injuries. It seems like the common thread here is that injuries are part due to the athletic trainer and part luck. I think coaching, scouting and the players themselves have something to do with it. Dallas has been relatively very healthy since 2003 when Parcells came into town. Before that with Campo, they were a mess with injuries. 2002 was ridiculous with injuries.

Parcells just scared players into not getting injured. Plus, he was smart enough to run practices so they would avoid injuries and do things like force the Linemen to wear knee braces. Phillips seems to have a good idea of when to rest players in practice. Plus, I think some players like a Flozell Adams, a guy that is naturally a humongous person, is less likely to get hurt.

London Fletcher-Baker is a short guy that never gets hurt and I think he's more of the exception than somebody like Flozell or Leonard Davis.

Lastly, I think FO.com and most statisticians get fooled by Romo since he really hasn't had your typical career. I guess he would project very similar to Kurt Warner, but the sample size of elite statistical QB's who came from a very small college, were not drafted and were on the bench their first 3.5 seasons is probably too small to get a projection that has any validity to it. So who knows just how he's going to do in the future, although I think he'll do a lot better than projected.

by Temo :: Fri, 08/13/2010 - 9:55am

It's not Romo I'm worried about, it's that offensive line that he's made look way better than they really are and is still getting older.

by chemical burn :: Fri, 08/13/2010 - 10:51am

Temo, I watched both the Cowboys preseason games and I'm curious what you think of the o-line performance thus far. To me they look really, really bad (especially last night) and the back-ups are borderline incompetent. OAK had a surprising high sack rate last year, but CIN was one of the worst in the league... I know it's preseason, but if that were the team I root for, I'd be very nervous...

by Temo :: Fri, 08/13/2010 - 10:39pm

I don't trust them.

I also watch the Jets on a regular basis, and lets just say that the difference in line play between the two teams is VERY striking. The line as a whole can behaves exactly as you expect a group of 350 lbs. men to play. They can open up big holes in the running game (and so far this preseason that's been lacking as well), but they lack any sort of mobility in pass protection.

by Richie3Jack (not verified) :: Fri, 08/13/2010 - 3:54pm

I thought the starting O-Line played well against Cincy. Ran the ball well. Protected Romo with no problem. The backups were not good though. Although Montrae Holland is a solid backup, he's just hurt right now, but would take over at guard. Alex Barron is probably a decent backup swing tackle. Center is the big issue because there's nobody behind Gurode, who I'm not nutty about anyway. I didn't watch the Oakland game last night, but Oakland made short work of the Cowboys last preseason as well.

by Temo :: Fri, 08/13/2010 - 10:40pm

They weren't bad, per se, but I thought Romo made them look better against Cincy.

by Sonofmer (not verified) :: Sat, 08/14/2010 - 3:57am

FO/Barnwell, Nice job on your preseason win predictions for 2009. Just a superb, bang-up job, you should be proud.

Maybe you should look at analyzing why your preseason predictions are so incompetent?

Here is a nice sample for you to start with:

FO 2009 preseason predictions:

Chicago Bears - 10.5 (Nice call)
Minnesota Vikings - 8.8 (Good call here)
Green Bay Packers - 7.4
Detroit Lions - 5.8

NFC South
Carolina Panthers - 8.3
New Orleans Saints - 7.8 (Didn't the Saints win the title?)
Tampa Bay Buccaneers - 7.2
Atlanta Falcons - 6.6

NFC East
New York Giants - 10.0 (Giants were powerhouses last year)
Philadelphia Eagles - 9.3
Dallas Cowboys - 8.0
Washington Redskins - 7.8

NFC West
Seattle Seahawks - 9.9
St. Louis Rams - 8.2 (Huh?)
San Franciso 49ers - 5.7
Arizona Cardinals - 5.6 (Nice Call)

AFC North
Pittsburgh Steelers - 9.6
Baltimore Ravens - 8.8
Cincinnati Bengals - 6.9
Cleveland Browns - 6.6

AFC South
Indianapolis Colts - 11.5
Jacksonville Jaguars - 10.2
Tennessee Titans - 9.3
Houston Texans - 6.9

AFC East
New England Patriots - 11.4
Miami Dolphins - 6.4
New York Jets - 6.2
Buffalo Bills - 5.3

AFC West
San Diego Chargers - 12.5
Kansas City Chiefs - 6.7
Oakland Raiders - 6.0
Denver Broncos - 4.9

by John Markson (not verified) :: Sat, 08/14/2010 - 11:54pm

Compared to the year before, it predicts direction of change 66% of the time (and for teams with the same record in '08 and '09 I gave the predictions credit if it was within half of the average change, which was 1.2).

However, if you use '08's data, the absolute difference in wins was 78 for the league, while those predictions were 82.6, so it was actually worse than just using last years records in that sense.

Furthermore, '08's numbers were within 2 wins 66% of the time, while the predictions were within 2 wins only 38% of the time.

If these trends hold up, it seems as though the FO predictions are good at predicting direction of change, but are weighted too heavily.

The real question is whether these predictions are better than vegas betting odds. My guess is that they aren't.

by thendcomes :: Sun, 08/15/2010 - 11:08pm