Dallas Morning News looks to Football Outsiders for answers

by sublimezg on Aug 21, 2008 11:17 AM CDT Comment 9 comments

Best idea DMN has come up with in some time. Football Outsiders is a great site that goes beyond the norm. Interesting to me what they came up with regarding Roy and his play last year. Also, it's fantastic to hear what Pac is capable of, I know we all think he can be great but this just goes to prove it. Now let's just hope he gets reinstated and stays that way. There's plenty more so go check it out.

http://cowboysblog.dallasnews.com/archives/2008/08/five-downs-with-football-outsiders-is-ro.html

Another user-created commentary provided by a BTB reader.

0 recs | Comment 9 comments

Story-email Email Printer Print

Comments

They Have Some Good Stuff

But they also whiff pretty badly. They predicted Dallas would win seven games in 06, six games last year, and only eight this year. That system also confidently predicted the 2005 Eagles to have the strongest indicators they’ve ever seen (they went 6-10), and so on.

The DVOA stuff is full of all sorts of dubious assumptions, too. It’s not to be taken too seriously, despite their posturing as deep thinking statisticians.

I will grant some of the simpler stuff they do has merit, like WR catch rates, and how opposing receivers fare against secondaries, which shines a light on how bad Roy Williams’ coverage is. Otherwise, it’s more smoke than fire.

by kindablue on Aug 21, 2008 6:08 PM CDT reply actions 0 recs

It's an imperfect art

And football projection will never be as clear-cut as baseball prophesying. But FO is the best in the business; DVOA seems to inexplicably favor some teams over others, but is otherwise a very accurate system. The Lewin projection system is money, the RB Speed Score system has merit, and their new WR Body-Types chart is very, very intriguing. Their DB assessments are the only tangible evidence we have of coverage aptitude, as is their analysis of stop rate, success rate, catch percentage, and countless other mind-twisting numbers I’ve wasted more time than I care to admit assessing.
They’re also top dog in fantasy projection, which has no actual football merit but shows you they have a knack for picking up regression, breakouts, and stability.
Frankly, I think all they need is a world-class draft guru (the role Kevin Goldstein fills for BP) and that site will become very, very popular.

Pretty clearly the first time he has ever looked at a web site. "And there are readers' comments at the bottom! And advertisements for products! And best of all, I won a free iPod!"
-Ken Tremendous on Mike Downey.

by BudLight on Aug 21, 2008 6:27 PM CDT up reply actions 0 recs

Imperfect Is an Understatement

That’s sort of like saying a field of cow manure doesn’t quite have the same aroma as fresh linens. When a system flops as badly as theirs does, it makes a mockery of what a competent prediction system should be.

And they are not the best in the business. Bill James came up with the simple and reliable Pythagorean method almost three decades ago, for baseball. It’s still a better predictor of who will win the Super Bowl than DVOA. The Pythagorean method holds every advantage over DVOA: It came first, it’s simpler, it’s more accurate, and is open to use by the public. That’s not even getting into the bad statistical basis of DVOA, but that’s a whole other story.

They do run the most aggressive campaign of self-congratulation I’ve ever seen amongst allegedly serious sports observers I’ve ever seen, including what is obviously a sickening amount of Internet sock puppetry. I’ll grant you they’re the best in that category.

by kindablue on Aug 21, 2008 7:39 PM CDT up reply actions 0 recs

The pythagorean system is very good

But as far as projection, it has its own imperfections because you cannot properly evaluate the changes a team has undergone in a given season. There is no individual analysis in the pythagorean formula. Now, if you want to involve WARP3, FIP, and all sorts of other things that will confuse everyone else, you’re adding so many new elements to the mixture that weren’t there before to create a more precise method of prediction.
My point? The two aren’t exclusive. And I’m sure that in the future, the necessary revisions to DVOA will be made. Which is why I continue to swear by them despite my shameless homerism and disgust for their continued Cowboys bias: They’re currently the one hope of football fans everywhere who pursue detailed statistical analysis.
Yeah, they’re cocky. Overly so, perhaps. Does it matter? Not really. Most internet phenoms are.

by BudLight on Aug 21, 2008 8:33 PM CDT up reply actions 0 recs

But Pythagorean Is Still Better than DVOA

Which is what we are talking about, and which refutes any claims that the FO analysis is the best system around. It clearly isn’t.

No one ever said the Pythagorean Method was perfect, just a significantly better tool than their flagship analysis.

The one hope of football fans for detailed statistical analysis? Please, don’t make me ill. They certainly go heavy on the detailed part, but a good statistician is more concerned with being accurate than being detailed. If FO cared about accuracy, they would use Pythagoras instead of DVOA.

by kindablue on Aug 21, 2008 10:03 PM CDT up reply actions 0 recs

i think the problem goes beyond that

its tough to use football stats b/c the season is short so the sampling size is small.

stats make sense in baseball b/c you have 162 games so trends end p being relevant. in other words lots of guys get hot for a month but by the end of the season the statistical leaders are pretty consistent. hot streaks and cold spells balance out.

in football its tough to come up with systems that use stats well (to predict analyze or explain) because the stats themselves arent that informative.

put another way, romo had 4 or 5 int’s vs buffalo. take away that game and his year looks MUCH MUCH better. the sampling size is small so each games matters too much.

by 325424 on Aug 21, 2008 10:14 PM CDT up reply actions 0 recs