What is the footballguys consensus on footballoutsiders.com? (1 Viewer)

[UCLAGIE]

http://www.footballoutsiders.com/2006/11/0...-rankings/4513/

Week 10 Power Rankings

Rk Team

1. BEARS

Projected record: 13-3 As we've mentioned in past weeks, major changes in DVOA can be hidden by the fact that our FOX articles only list the rankings, and not the actual ratings. Chicago was way ahead of everybody else a week ago. The Bears are still number one, but these top four teams are so tightly packed together that the difference between first and fourth isn't really significant. NEXT: at NYG

2. CHARGERS

Projected record: 13-3 The Chargers had to be celebrating when the Colts beat the Patriots this week. Obviously, we have no idea which playoff seed the Chargers will get, or even if they will make the playoffs -- the Merriman suspension and injuries in the front seven probably mean that projected record is a bit too optimistic. But if the Chargers are one of the final four teams standing, where would they rather play the AFC Championship game: in freezing Foxboro, Baltimore, or Denver? Or in the climate-controlled confines of the RCA dome, where they upset the Colts just a year ago? NEXT: at CIN

3. GIANTS

Projected record: 11-5 Former FO writer Al Bogdan explains the Giants near-loss to Houston: "The Giants can't get any pass rush going without Michael Strahan and Osi Umenyiora. And without Plaxico Burress, there isn't anyone who can catch Eli Manning's patented overthrows. He had one on a third-and-long to Michael Jennings that Burress (6'5") usually has a shot at coming down with, but Jennings (5'11") has no chance to catch. The Giants still have no problem running the ball, so eventually they just gave up on Eli throwing deep without Burress to serve as a backboard, and mixed up short passes with a lot of Tiki Barber and Brandon Jacobs." NEXT: vs. CHI

4. EAGLES

Projected record: 10-6 You know, now that it seems that nearly every team in the NFC is 4-4, our rating for the Eagles doesn't look quite so out of place. Since there are only five teams in the NFC with winning records, some 4-4 or 3-5 team is going to end up in the playoffs. I have no idea how anybody could argue that Philadelphia is not the best of those teams. NEXT: vs. WAS

5. RAVENS

Projected record: 12-4 Colts fans may ask: Why are the equally one-dimensional Ravens ranked higher than the Colts? You may notice that the Ravens actually rank higher on offense than the Colts do on defense. But more importantly, the Ravens are far superior to the Colts in that third dimension of football: special teams. Baltimore is above-average in all five parts of special teams that we measure: kickoffs, kick returns, punts, punt returns, and field goals. NEXT: at TEN

6. JAGUARS

Projected record: 10-5-1 If Maurice Jones-Drew had a sister, and she married London Fletcher-Baker of Buffalo, would their kids be Jones-Drew-Fletcher-Bakers, or Fletcher-Baker-Jones-Drews? NEXT: vs. HOU

7. COLTS

Projected record: 13-3 Wait, I didn't write enough about the Colts up top? OK, here's another tidbit: Last year, the Colts ranked second in Adjusted Sack Rate, which measures sacks per pass play adjusted for situation and opponent. This year, the Colts are 25th. However, they could still match last year's sack total if they could petition the league to change their next four opponents to "Oakland." NEXT: vs. BUF

8. PATRIOTS

Projected record: 12-4 You know, I think they would have won that game if they had just re-signed Adam Vinatieri. NEXT: vs. NYJ

9. BRONCOS

Projected record: 11-5 Early in the season, on the Football Outsiders FOX blog, I wrote that the Broncos defense couldn't keep up it's amazing early performance, but the offense would improve to balance things out. By conventional stats, it looks like the Denver defense has gotten worse, but in DVOA they have the exact same ranking: 11th. Meanwhile, the Denver offense has moved up from 17th to eighth. By the way, a belated thanks to Jordy Singer and Jason Beattie for helping with the local Denver humor in last week's comment. NEXT: at OAK

10. COWBOYS

Projected record: 8-8 I must admit that I'm a little surprised by the projection system's pessimism about the Cowboys, but it doesn't like that both the offense and defense get worse in the second half of close games. Then again, it doesn't realize that the new quarterback doesn't have a giant fork in his back, so I would lean towards 9-7 or 10-6 instead. NEXT: at ARI

11. CHIEFS

Projected record: 10-5-1 If you have a lot of Chiefs on your fantasy team, you may want to look into some trades -- unless the Chiefs you have are the Chiefs defense. In the last eight weeks, Kansas City has the hardest offensive schedule in the league, by a huge margin. They also have the easiest defensive schedule in the league. They play Oakland twice, Miami, Cleveland, Baltimore, and finish the season with Jacksonville -- all teams that are far better on defense than on offense. San Diego and Denver, of course, can play some defense too. NEXT: at MIA

12. STEELERS

Projected record: 6-10 With the star-crossed Steelers now completely out of it, Casey Hampton can return to his other job as lead singer of Gnarls Barkley. NEXT: vs. NO

13. SAINTS

Projected record: 10-6 Everybody knows that the Saints have been better running the ball up the middle rather than around the ends this season. But did you know that the defense is also better up the middle? FO's Adjusted Line Yards stats (see here) rank the Saints 29th stopping runs around left end or right end, but fourth stopping runs listed as going behind a guard or up the middle. NEXT: at PIT

14. RAMS

Projected record: 8-8 Alex Barron is a talented young lineman – check out this article Michael David Smith wrote about him last year – but he just can't sit still. Barron led the league with 18 false starts last year, and he blew a fourth-quarter drive that would have tied things up against Kansas City by false-starting twice in a row and turning second-and-1 on the Kansas City 26 into second-and-11 on the Kansas City 36. Dude, you must chill. I have hidden your keys. NEXT: at SEA

15. BENGALS

Projected record: 7-9 Am I the only one who thinks the whole "Chad Johnson is complaining about his catches" story is a tempest in a teapot? It isn't like Ocho Cinco called a press conference so that he could publicly berate Marvin Lewis and Carson Palmer. A reporter went to Johnson after the game, prodding him for a quote, and trying to get him to say something controversial, and Johnson was depressed because the Bengals had just lost to an important division rival. What star receiver doesn't think he could help his team win if he could just get the ball more? NEXT: vs. SD

16. PANTHERS

Projected record: 8-8 Things the midseason projection system likes about Carolina: They have a strong offense in the first quarter, which helps a team dictate the pace of a game, and they've improved on both offense and defense since the first couple games of the year. Things the midseason projection system does not like about Carolina: Both offense and defense get worse in the second half of close games, only San Francisco and San Diego have worse defense in the red zone, and the Panthers have one of the five hardest remaining schedules in football. NEXT: vs. TB

17. VIKINGS

Projected record: 9-7 For 12 solid months now, the Vikings have been winning with defense while offense gets the credit. I think at this point, people are finally getting the hint. Brad Johnson is old, there are no standout receivers, and Steve Hutchinson can't stop you from beating up on the Vikings' tackles. Meanwhile, the defense has been awesome against everyone but the Patriots. NEXT: vs. GB

18. PACKERS

Projected record: 6-10 This rating sure seems high, doesn't it? The Packers benefit from the switch to the system that considers recent games more than early games, since their biggest loss of the season came in Week 1. Notice that the "Full Year" ranking switches the Packers and Falcons. The Packers also get a bit of a boost because they've had poor luck with fumble recovery. NEXT: at MIN

19. REDSKINS

Projected record: 7-9 Although he was active on Sunday, T.J. Duckett might as well not be on the roster. He hasn't touched the ball since the Week 2 Dallas game. What was the point of giving up a third-round draft pick for this guy? If Clinton Portis is looking to play another character, how about dressing up in T.J. Duckett's uniform and calling himself the Invisible Man? NEXT: at PHI

20. FALCONS

Projected record: 9-7 Since I know many people can't remember these comments from week to week, I'll repeat from last week: Atlanta would rank a few places higher if we didn't count the Monday Night Massacre in New Orleans. Of course, losing to Detroit isn't quite so easy to rationalize. NEXT: vs. CLE

21. DOLPHINS

Projected record: 5-11 That upset of Chicago was a tantalizing peek at the team that Dolphins fans thought they were getting. But the long-term picture still isn't very good. The Dolphins are locked into Daunte Culpepper's contract and have to hope he can get it together. The front seven isn't getting any younger, and one good game by Yeremiah Bell isn't going to solve the problems in the secondary. Trying to fill holes with young talent? Don't forget there's no second-round pick next year. NEXT: vs. KC

22. BILLS

Projected record: 6-10 Is there anybody outside of upstate New York actually paying attention to the Bills? They might as well change the uniform colors to beige and off-white. They aren't particularly good at anything except special teams, and they aren't particularly terrible in any way that's easy to make fun of. If a team beats Green Bay 24-10 at home, does it make a sound? NEXT: at IND

23. SEAHAWKS

Projected record: 9-7 If you've attended Seahawks training camp over the last two years, you've seen backup quarterback Seneca Wallace throw perfect rainbows to unheralded receiver D.J. Hackett. Little surprise that Hackett has become a primary target for Wallace as the Seahawks continue to deal with Matt Hasselbeck's absence. Last year, Hackett was the most valuable receiver in the NFL thrown less than 50 passes, according to our DPAR stats. Our midseason projections expect the Seahawks to play more like defending conference champions in the second half, and if they sweep the season series with the Rams, they should be the team to break the Super Bowl Loser's Curse. NEXT: vs. STL

24. BROWNS

Projected record: 5-11 If Pittsburgh is the most surprising team in the NFL this year, their division rivals in Cleveland are the least surprising team. Perhaps the special teams have been better than people thought, and nobody really knew if Kellen Winslow would be able to reach his potential after two years of injuries (he could, and fast). But otherwise, this Cleveland season has pretty much played out exactly as expected. The young players who were supposed to be gaining experience are, in fact, gaining experience, the injury-prone Gary Baxter once again is injured, and the target date for this team is still 2007. NEXT: at ATL

25. LIONS

Projected record: 5-11 Kevin Jones and Mike Martz seem to have come to an understanding: Martz won't ask Jones to run Marshall Faulk-style deep routes, the way he did early in the season, and Jones will stop dancing around in the backfield and just run the play as Martz drew it up. It's working: Jones has 323 rushing yards and 123 receiving yards the last three weeks, and the Lions are finally starting to resemble a professional football team. NEXT: vs. SF

26. JETS 4-4

Projected record: 6-10 The Jets' offense has slowed down in October despite poor competition, and it has trouble establishing itself early in games, ranking 26th in first-quarter DVOA. Combine those two trends with a few other indicators, and our second-half statistical projection expects the Jets offense to collapse, taking their season with it. Still, 6-10 would be a lot better than what people expected in August. NEXT: at NE

27. BUCCANEERS

Projected record: 5-11 Things you don't want to hear while watching your favorite team: "Philip Buchanon now in at corner for the injured Juran Bolden." NEXT: at CAR

28. TEXANS 2-6

Projected record: 5-11 While much-heralded 49ers rookie Vernon Davis sits on the sidelines with an injury, Texans rookie Owen Daniels is putting together a nice little rookie season, with 217 yards and five touchdowns. Too bad his blocking is horrible -- and by the way, veteran Jeb Putzier isn't any better. NEXT: at JAC

29. 49ERS

Projected record: 6-10 The 49ers were on pace to set the single-season record for points allowed through their first seven games. Then they gave up just three points on Sunday. That says more about Minnesota's current offense than any real turnaround on the San Francisco side, but let's give props to linebacker Brandon Moore, who subbed for Jeff Ulbrich and racked up a career-high 14 tackles. NEXT: at DET

30. RAIDERS

Projected record: 4-12 It is just about inconceivable that a team coached by perhaps the best left tackle in NFL history in Art Shell, and with another Hall of Fame O-lineman in position coach Jackie Slater, can have an offensive line this bad. Note to offensive coordinator Tom "Basil Fawlty" Walsh: Cut out the seven-step drops before you get Andrew Walter killed. NEXT: vs. DEN

31. CARDINALS

Projected record: 3-13 The bye week was a welcome diversion for the Cardinals. Dennis Green could pretend to outcoach an opponent, Edgerrin James could pretend to run past the line of scrimmage, and Matt Leinart could pretend he was back at USC, where he had actual blocking. NEXT: vs. DAL

32. TITANS 2-6

Projected record: 3-13 Hey, remember that 178-yard game that Travis Henry had against Washington? Some people thought it was a fluke, but it turns out it was … a massive fluke. Take out the games against Washington and Indinapolis, and Henry is averaging 2.8 yards per carry. Seriously, Tennessee, stop futzing around with Henry – he's just not a starting-caliber back. Let's get LenDale White some experience already. NEXT: vs. BAL

http://msn.foxsports.com/nfl/story/6147120

 

[Mootej]

I'd rate Indy up at least 2 places at #5 over Baltimore and the Jaguars. Read the DVOA and the way FBO manually moved Indy up last year. HAHA, don't trust your instinct. This year, they let the stats and the computer rank stand.

I agree, with them, Indy hasn't looked that good. They have squeaked a couple of games out. They did look good IMHO against NEP but Harrison ruled that game. Maybe #1SDC, #2 Philly, #3 Colts but certainly Colts not lower than #5.

 

[thatguy]

I know there are a number of guys here who are very high on FO. Maurile and SSOG are the two that come to mind right off hand.

I for one am a bit skeptical of their model's actual predictive power.

Basically, they take into account every single play, accounting for opponent and game situation, and compare it to the league average in similar situations. They take all of the results, and through a series of computations, come up with a metric that they call DVOA, which is either positive (better than average) or negative (worse than average). They do this for offense, defense, and special teams, and compile the DVOA of each play for each team to come up with their total DVOA.

They continually tweak the model to increase the correlation between DVOA and winning. So, it's basically a backwards process. They see how close the correlation is between DVOA and winning in the current model, tweak something here or there, see if the correlation factor increases, or decreases, and so on. Last season when they had the Colts ranked so low despite popular opinion that they were hands down the NFL's best team, they tweaked the effect that SOS had on the model, and this "fixed" it somewhat. (They made no such tweak this year, as they are standing by their ranking of the Colts and saying that everyone else is wrong.)

They also compute similar numbers for offensive skill players, as well as O Lines and D Lines.

Anyway, that's my basic understanding of what they do, but I haven't put a whole lot of time into fully understanding the exact nature of the process. I'm sure Maurile or SSOG can give a much better explanation.

It's a complicated model for sure, but I really think the idea is too simplistic. No matter how fancy you get in your statistical analysis, football, and a given team's success or lack thereof, just can't be explained by mere numerical models. There is so much going on during the course of a game that no statistical model simply can possibly account for.

Still, I think it's a better metric than the more commonly uses metrics like YPG, PPG, etc.

 

[David Dodds]

Falcons at 20 and Seahawks at 23 aso seem way off. I think their model needs more work.

 

[djcolts]

I think they rely too heavily on how a team beats the bad teams in their ratings, which is why the Bears are #1, Chargers are #2, and the Eagles are #4. Those 3 teams have combined to beat ONE team over .500 this year.

 

[Black&Gold]

When you can't stop the run, you won't win a championship. Thus the Colts are at #7.

 

[KnowledgeReignsSupreme]

I've been reading their sites for years. I like their out of the box thinking.

 

[KnowledgeReignsSupreme]

Nice comments for Jacksonville.

 

[ConstruxBoy]

I've been reading their sites for years. I like their out of the box thinking.

 

[djcolts]

When you can't stop the run, you won't win a championship. Thus the Colts are at #7.

Why aren't they #6? Or #8? Or #3? Or #15? Based on "logic" - if your first sentece is correct, then the Colts can be ranked anywhere from #2 to #32. Therefore, your post is flawed.

 

[Black&Gold]

When you can't stop the run, you won't win a championship. Thus the Colts are at #7.

Why aren't they #6? Or #8? Or #3? Or #15? Based on "logic" - if your first sentece is correct, then the Colts can be ranked anywhere from #2 to #32. Therefore, your post is flawed.

No, they do have the best offense in the league. So they will beat some people. Actually, they will glide through the regular season and may even win a playoff game. But somone will control the clock, run it down their throat, and win the game. Sound familiar? So the Colts are good enough to be #7, but there are about 6 teams that will be them in the playoffs.

 

[guderian]

I've been reading their sites for years. I like their out of the box thinking.

I like what they do. It's not without flaws, but they try and adjust for the small sample sizes in the NFL (only 16 games) by looking at individual plays and comparing those to the league average. This helps in comparing players that are frequently in the game during 3rd down situations, for example. They've also made some pretty good predictions over the past few years with their surprise teams/players. The problem is that too many people try and force their stats into a fantasy football context.BTW, I also like their out of the box thinking and unusual predictions. Too often football sites will have fairly narrow differences with other sites on players/teams. FO will crank through their stats and come up with some very unique conclusions.

 

[AB in DC]

Still, I think it's a better metric than the more commonly uses metrics like YPG, PPG, etc.

Agreed -- there are a lot of problems with their measures, but there's nothing better out there. And compared to some of absurd power rankings I see on other sites, theirs will at least include some reasonable justification instead of snide commentary.

 

[hector31]

It's an interesting site. It's different. Worth poking around, for sure. A lot of intelligent people there. At times some of their guys (really just one guy) come off a little arrogantly, but given how much work they've put into the baby, I guess I can see that. I'm not justifying it, just trying to understand it.

As for their team ranks, it's true that part of the problem is that you get too much credit for blowouts. The Colts get hurt by their "ordinary" wins over the Titans, Jets and some others. But if I have to bet my estate on one team winning it all in 2006, give me the best offense and the best QB, in a second.

 

[ConstruxBoy]

I've been reading their sites for years. I like their out of the box thinking.

I like what they do. It's not without flaws, but they try and adjust for the small sample sizes in the NFL (only 16 games) by looking at individual plays and comparing those to the league average. This helps in comparing players that are frequently in the game during 3rd down situations, for example. They've also made some pretty good predictions over the past few years with their surprise teams/players. The problem is that too many people try and force their stats into a fantasy football context.BTW, I also like their out of the box thinking and unusual predictions. Too often football sites will have fairly narrow differences with other sites on players/teams. FO will crank through their stats and come up with some very unique conclusions.

Yep. I used their KUBIAK rankings to augment some of my own projections this pre-season and one of the interesting things I saw was Neil Rackers ranked very low. Now I already knew that Kickers are very up and down each year, but it was nice to see a site really rank him where I thought he could finish with a bunch of other PKs above him like this Robbie Gould guy.

 

[mbuehner]

Here are my ingredients for championship football in order of importance:

1.defense

2.running game

3.being able to score when 'necessary' (eg playoff experience, 2-minute drill play)

4.red-zone success in general

-The Bears cant run the ball well enough right now, and their ability to turn up the volume offensively when they are in a bind is a major question mark.

-Colts dont have the defense to do it, and im not sold on their running game.

-In my opinion, the Patriots have the best chance of winning the superbowl as of today, using my criteria.

 

[SSOG]

It's a complicated model for sure, but I really think the idea is too simplistic. No matter how fancy you get in your statistical analysis, football, and a given team's success or lack thereof, just can't be explained by mere numerical models. There is so much going on during the course of a game that no statistical model simply can possibly account for.

Of course a lot of that is true- a statistical model isn't going to understand that your starting QB was hurt, or that you were resting your starters for the postseason because your seeding was clinched... but on the other hand, this statistical model every year winds up kicking the hindquarters of every other projection out there. You could get 30 football experts, guys who are paid to do nothing but analyze football, really intelligent guys like Ron Jaworski. You could put them in a room and have them hash over all of the variables, and then come up with a projection... but DVOA would still routinely outperform them.For two years, I tracked DVOA vs. every other standard power ranking I could find in terms of how they predicted individual games *AS WELL AS* season-ending standings. I used actual season-ending rankings for the latter part, which I understand gives me very flawed numbers (since I'm basically comparing rankings to themselves, and a power ranker who never changed a single thing all season would come out on top), but despite my admittedly flawed method, DVOA consistantly came out on top of every "expert" out there.

Falcons at 20 and Seahawks at 23 aso seem way off. I think their model needs more work.

Their model does need work- they'll be the first to admit it, which is really refreshing. They're constantly finding places where their prediction is broken, or just seems wrong, and they're constantly testing out new tweaks and changes to see if they actually improve the predictive quality of DVOA. Still, even if those two rankings are way off (not saying they are, but IF they are)... two bad rankings are an outlier, not an indictment of the model itself. Every model is going to produce outliers, and theirs is no exception. They frequently will flag certain data points that they think are probably outliers and will explain the reasoning behind why they think it's an outlier.Still... don't sell DVOA short. Some examples of some of the "outliers" DVOA has spit out in the past:

1) Prior to the 2003 season, DVOA predicted that the WOEFUL San Diego Chargers, who were coming off of a 4-12 season and had their sportswriters joking that they were the worst team on their own schedule, would be a playoff team. DVOA didn't just predict that they'd be a playoff team, DVOA predicted that the Chargers would have one of the best offenses in the entire NFL. I'm pretty sure that DVOA was the only entity in the known universe that predicted a playoff berth for San Diego that year.

2) Last year, DVOA kept predicting that the 6-2 Atlanta Falcons were a below average team, which resulted in REAMS of hate-mail from Atlanta fans. Atlanta finished the season 2-6.

3) Last year, Eli Manning started with a 14:5 TD:INT ratio, and Football Outsiders called it a statistical aberration of HISTORIC proportions. Over the rest of the season, including Playoffs, Manning threw 10 TDs vs. 15 INTs.

Now, they've had some predictions miss, too, and obviously this isn't a complete list of all their good predictions, but the point is that just because a prediction seems way out of line doesn't mean it's wrong- it could just be DVOA picking up on a trend before the rest of the world catches on to it.

One last point in DVOA's defense- every year, King Kaufman tracks preseason predictions and compares that to actual playoff teams. For two years running, Football Outsiders has taken home the preseason predictions crown.

I think they rely too heavily on how a team beats the bad teams in their ratings, which is why the Bears are #1, Chargers are #2, and the Eagles are #4. Those 3 teams have combined to beat ONE team over .500 this year.

Read this article. It goes on to detail how the biggest indicator of which teams will do well in the postseason isn't which teams do the best against good teams, it's which teams do the best against bad teams. A team that plays the 15 worst teams in football and beats them all by 50, and then loses by 3 to the only good team they play all year is more likely to win the superbowl than a team that plays the 15 best teams in football and beats them all by 3, but then loses by 30 to the worst team in the league.Also, like I said, they're constantly tweaking their projections. Aaron has tried 4 or 5 times now to make blowout wins count for less in the formula, but every time he's tried, it has actually REDUCED the predictive power of DVOA. Which means that those blowout wins against crappy teams tell us an awful lot about how good a team really is, after all.

As you can all probably tell by now, I'm a huge Football Outsiders junkie. I use a lot of their statistics to gain a decisive fantasy edge against my league. For instance, looking at their offensive line numbers showed that no team's rushing success was MORE attributable to its offensive line than Indy, and no team had a worse offensive line than Arizona. The obvious conclusion, then, was that Edgerrin James was going to be one of the most spectacular flameouts we've seen in years. Their DVOA vs. #1 WRs, #2 WRs, and TEs is often a pivotal tiebreaker when setting my starting lineup. I even use their DVOA rankings when picking teams in eliminator pools. They've definitely steered me right more often than not.

 

[Aaronstory]

I've been reading their sites for years. I like their out of the box thinking.

 

[zoonation]

Can someone explain the Pittsburgh ranking for me? TIA.

 

[AB in DC]

Can someone explain the Pittsburgh ranking for me? TIA.

They were a pretty good team with Charlie Batch. The blowout of 5-3 KC offsets a couple close losses.They were unlucky in two losses -- Oakland (a goal line interception is almost NEVER returned for a 100 yard TD) and Atlanta (coin flip). Might add the Cincy game too, though I forget the details. Based on the stats, all three were virtually ties.I don't think anyone disagrees that they have a top-ten, maybe top-five defense.

 

[AB in DC]

Also, by the numbers, they've played the sixth-toughest schedule of all teams, including three very good teams (SD, Denver, Jax) and only one very bad team (Oakland).

Or four good teams (counting KC) and two bad teams (counting Miami).

 

[thatguy]

SSOG, did you check to see if there was any statistical signifcance in the difference between DVOA and the rest of the herd? Because if not, I'm not sure it's very meaningful (I realize that for such a small sample size it would be very difficult to prove anything in a statistically meaningful way, but that is another one of the inherent flaws I see in what they're doing - they will never really know if they're more right than the rest, or just more lucky). Also - and I'm not saying they do this - I think it would be funny for a site that makes it's living with hardcore statistical analysis to use something with no statistical significance whatsoever as proof of its successes.

 

[thatguy]

I guess what I'm saying is that only time will tell how good their model really is. Outdoing the "competition" for a couple years just isn't very telling. If they outdo them for the next 5 or 10 years, then I think we can say they're onto someting. Also, just because they outdo the so-called "experts" does not mean a whole lot. A lot of the experts are oftentimes way off base.

As for my saying that their system doesn't take everything into account, I'm not just referring to injuries or resting starters.

Imagine a game between team A and team B. Team A jumps out to an early 21-0 lead. That will most likely net them a very positive DVOA and net team B a very negative DVOA. But team A's coach gets conservative and resorts to the run. Team B is expecting the run, so they sell out to stop it. Team A has a lot of stalled drives, which is fine because they're sitting on a lead. Because most of these runs will likely be worse than the league average due to Team B's adjustments, these drives will net Team A a lot of negative DVOA. On the flip side, Team B's defense will be doing better than the league average, so they will get a lot of positive DVOA. To me, this is not an accurate description of what's going on, or how "good" each team is. It is measuring success on each given play, sure, but it fails to take into account coaching style and game texture. Now, Team A also decides late in the game that they will get conservative on defense, because their lead still stands, and they drop 8 guys into deep zones and rush 3, allowing Team B to dink and dunk its way down the field for some garbage points. Let's say the final score is 31-17. In this garbage time, Team A's defense doesn't care about the consistent 8-12 yard gains, because they know Team B is going to have to score a whole lot quicker than that to get back in the game, yet they are consistently worse than the league average on every single play, and therefore net negative DVOA. Team B's offense, which has done nothing thus far, consistently outperforms the league average due to what Team A is giving them, and nets a positive DVOA. The bottom line is that in this game, Team A probably still gets a positive DVOA, Team B a negative DVOA, but it is still not an accurate description of how good either team is. I know that you can say, well over the course of the season, this will even itself out for all teams, but that's just not true. Some coaches are far more conservative. Some teams consistently get into close games where every play really does matter. I don't know, what I'm saying is that no numerical model can take these intangibles into account so I just don't see how any numerical model can ever be an incredibly good indicator of how good a team is.

 

[thatguy]

And perhaps this argument has already been coverage over at FO. If so, I apologize, as I go there sparingly and never read the commentary. Either way, I don't think there is a good good answer to this problem - at least not one that I would find satisfactory.

 

[thatguy]

I'd also be curious to hear some of the "ballsy" picks or predictions that FO has made that no one else saw coming.

 

[AB in DC]

But team A's coach gets conservative and resorts to the run. Team B is expecting the run, so they sell out to stop it. Team A has a lot of stalled drives, which is fine because they're sitting on a lead. Because most of these runs will likely be worse than the league average due to Team B's adjustments, these drives will net Team A a lot of negative DVOA.

Not true. Team A's run will be compared to all other teams who run the ball with an early 21 point lead, not to all runs, period. So if there's a negative hit there, it'll probably be pretty small.

 

[thatguy]

But team A's coach gets conservative and resorts to the run. Team B is expecting the run, so they sell out to stop it. Team A has a lot of stalled drives, which is fine because they're sitting on a lead. Because most of these runs will likely be worse than the league average due to Team B's adjustments, these drives will net Team A a lot of negative DVOA.

Not true. Team A's run will be compared to all other teams who run the ball with an early 21 point lead, not to all runs, period. So if there's a negative hit there, it'll probably be pretty small.

Ah, thanks for clearing that up. I suppose then the one thing it still won't take into account is coaching philosophy, but that makes me much happier.

 

[MLBrandow]

My opinion is that this is nothing but bathroom reading material.

It has little substance and a lot of quips.

What i do like about it versus per say ESPN's power rankings is that it's not continually loaded with "Obviously"s. They need a damn editor.

 

[guderian]

I guess what I'm saying is that only time will tell how good their model really is. Outdoing the "competition" for a couple years just isn't very telling. If they outdo them for the next 5 or 10 years, then I think we can say they're onto someting. Also, just because they outdo the so-called "experts" does not mean a whole lot. A lot of the experts are oftentimes way off base.As for my saying that their system doesn't take everything into account, I'm not just referring to injuries or resting starters.

The factors that you cited as negatives are what I like about their system--their system doesn't try to adjust for these subjective factors. If it did, they would probably subjectively move up Indy, for instance to match other power rankings and not look off base. As a result, when they have teams like Indy or the Falcons ranked much lower than you'd think that spurs you to think about those teams from a different perspective. Then you can dig into it and find out what factors are causing this discrepancy. Maybe you agree with it, or maybe you don't; but it's information that you don't typically get anywhere else. So, I don't care if their model is a better predictor than other models, but I do appreciate that it generates outside the box conclusions that you rarely get anywhere else--afterall where else are you going to see Indy ranked #7 and Atlanta ranked in the 20s?

 

[SSOG]

But team A's coach gets conservative and resorts to the run. Team B is expecting the run, so they sell out to stop it. Team A has a lot of stalled drives, which is fine because they're sitting on a lead. Because most of these runs will likely be worse than the league average due to Team B's adjustments, these drives will net Team A a lot of negative DVOA.

Not true. Team A's run will be compared to all other teams who run the ball with an early 21 point lead, not to all runs, period. So if there's a negative hit there, it'll probably be pretty small.

Ah, thanks for clearing that up. I suppose then the one thing it still won't take into account is coaching philosophy, but that makes me much happier.

No way should they adjust for coaching philosophy. Good plays are good plays whether your philosophy is conservative or aggressive. Likewise with bad plays. I mean, if a team throws a 90 yard TD-pass, would it make any difference whether their head coach liked to pass 50% of the time or 75% of the time?DVOA is very results-oriented. Your reasoning behind making moves is irrelevant, it's only the outcome that matters.

I also like the fact that DVOA is an actual value, rather than a strict ranking. I mean, in most power rankings, it looks like the difference between the #3 and #4 team is just as great as the difference between the #4 and #5 teams, but DVOA clearly shows tiering. Comparing DVOA to Power Rankings is sort of like comparing tiered rankings and projections to just straight up positional rankings in fantasy football- DVOA just gives a *LOT* more information, and more information is almost always a good thing.

As for some of their more outside-the-box predictions...

http://www.footballoutsiders.com/2004/07/29/ramblings/238/

http://www.footballoutsiders.com/2005/07/29/ramblings/2742/

http://www.footballoutsiders.com/2006/07/28/ramblings/4049/

Those are their year-end highlights from each of the last three seasons, featuring some of the most notable and memorable articles from the year-in-review. On those pages, you should find some links to several midseason reports and preseason previews, as well as a prediction review after the season was over.

Last season, DVOA picked the Patriots, Steelers, Colts, and Chiefs to win their divisions, and the Broncos and Bengals to be wildcards. Instead, the Steelers lost Roethlisberger to injury for several games and finished with the wildcard behind the Bengals, the Broncos won their division, and the Chiefs became probably the best team to miss the playoffs in the past decade (yes, better than last year's Chargers).

In the NFC, it wasn't as accurate, but it predicted the Eagles, Lions, Bucs, and Seahawks winning their divisions, and the Panthers and Cowboys as Wildcards. It totally whiffed on the Bears and Lions (right after Grossman was hurt, nobody was predicting anything from the Bears), and obviously the Eagles completely fell apart, but the rest of the picks were pretty solid. Tampa Bay and Seattle both hit, Carolina really was a wildcard, and the Cowboys were at least fighting for that last wildcard until the very end of the season.

Edit: Here's an article they wrote for the New York times discussing DVOA's surprising predictions for the 2004 San Diego Chargers, which also pointed out that there would be huge improvements on offense for the Seahawks and Panthers, and that both Delhomme and Hasselbeck would be much improved last year. Both QBs, in turn, set career highs in YPA. The Panthers saw their scoring rise from 13th in the NFL to 8th, and the Seahawks went from 12th to 1st.

 

[gheemony]

This isn't very helpful, but I have to say that I love footballoutsiders.com. Even better is their annual book, Pro Football Prospectus.

I now understand why I constantly find SSOG's analysis to be helpful - clearly open-minded and thoughtful.

Speaking of which, it would be great if we tag the FO-fans in the Shark Pool so I could sort some threads to ignore the non-FO-fans.

 

[The Beezer]

This isn't very helpful, but I have to say that I love footballoutsiders.com. Even better is their annual book, Pro Football Prospectus.I now understand why I constantly find SSOG's analysis to be helpful - clearly open-minded and thoughtful. Speaking of which, it would be great if we tag the FO-fans in the Shark Pool so I could sort some threads to ignore the non-FO-fans.

I read the Prospectus on the way to and from Canton for the induction ceremony in August. By the time I got home, I felt 10 times smarter about football than when I had left (and then drafted like it all month). Ignore them at your peril.

 

[thatguy]

This isn't very helpful, but I have to say that I love footballoutsiders.com. Even better is their annual book, Pro Football Prospectus.

I now understand why I constantly find SSOG's analysis to be helpful - clearly open-minded and thoughtful.

Speaking of which, it would be great if we tag the FO-fans in the Shark Pool so I could sort some threads to ignore the non-FO-fans.

I strongly suggest you ignore anyone who questions the usefullness of FO and its DVOA. We're clearly closed-minded, thoughtless morons who offer nothing of value to any discussion.

 

[SSOG]

This isn't very helpful, but I have to say that I love footballoutsiders.com. Even better is their annual book, Pro Football Prospectus.

I now understand why I constantly find SSOG's analysis to be helpful - clearly open-minded and thoughtful.

Speaking of which, it would be great if we tag the FO-fans in the Shark Pool so I could sort some threads to ignore the non-FO-fans.

I strongly suggest you ignore anyone who questions the usefullness of FO and its DVOA. We're clearly closed-minded, thoughtless morons who offer nothing of value to any discussion.

I don't ignore people who question DVOA. In fact, *I* question DVOA all the time. Aaron, DVOA's creator, questions DVOA all the time. I think it's important, with any model, that you never become satisfied with it, and that you always question it and look for flaws. For instance, back in 2003, there was a chicken-or-the-egg arguement about Tennessee's run defense where they led the league in conventional categories, but DVOA claimed they were mediocre and only led the league because they faced so few running plays. I always liked to argue that maybe they faced so few running plays because they were great against the run, and then after a while they faced so many passes that they just played pass all the time, and that's why individual runs were more successful but teams were refusing to run on them consistantly. I still think that's a big flaw in DVOA- it doesn't assume that, on the whole, coaches are intelligent human beings. I think if one coach refuses to run on a mediocre defense, that's a bad game plan... but if the ENTIRE LEAGUE is refusing to run on a mediocre defense, then maybe the coaches know something that DVOA doesn't and that that running defense isn't so mediocre after all.Now, with all that said, I usually ignore anyone who dismisses anything FO says out of hand because they're "stat geeks"... because if someone isn't willing to give the least little bit of credence to non-conventional statistics, that person and I aren't going to have very much to talk about. Doubt DVOA all you want, but discount it at your own risk.

 

[thatguy]

This isn't very helpful, but I have to say that I love footballoutsiders.com. Even better is their annual book, Pro Football Prospectus.

I now understand why I constantly find SSOG's analysis to be helpful - clearly open-minded and thoughtful.

Speaking of which, it would be great if we tag the FO-fans in the Shark Pool so I could sort some threads to ignore the non-FO-fans.

I strongly suggest you ignore anyone who questions the usefullness of FO and its DVOA. We're clearly closed-minded, thoughtless morons who offer nothing of value to any discussion.

I don't ignore people who question DVOA. In fact, *I* question DVOA all the time. Aaron, DVOA's creator, questions DVOA all the time. I think it's important, with any model, that you never become satisfied with it, and that you always question it and look for flaws. For instance, back in 2003, there was a chicken-or-the-egg arguement about Tennessee's run defense where they led the league in conventional categories, but DVOA claimed they were mediocre and only led the league because they faced so few running plays. I always liked to argue that maybe they faced so few running plays because they were great against the run, and then after a while they faced so many passes that they just played pass all the time, and that's why individual runs were more successful but teams were refusing to run on them consistantly. I still think that's a big flaw in DVOA- it doesn't assume that, on the whole, coaches are intelligent human beings. I think if one coach refuses to run on a mediocre defense, that's a bad game plan... but if the ENTIRE LEAGUE is refusing to run on a mediocre defense, then maybe the coaches know something that DVOA doesn't and that that running defense isn't so mediocre after all.Now, with all that said, I usually ignore anyone who dismisses anything FO says out of hand because they're "stat geeks"... because if someone isn't willing to give the least little bit of credence to non-conventional statistics, that person and I aren't going to have very much to talk about. Doubt DVOA all you want, but discount it at your own risk.

My comments were not directed at you, but I think you know that. As for DVOA, I don't discount it entirely. I said already that I consider it a better metric than the traditional ones that we use to measure success. However, I do have my doubts as to its usefullness. You have given some examples of its predictive ability. But until they show that they are consistently better than the next best thing (as I said for 5 or 10 years or more) I'll have my doubts. And if DVOA has no predictive ability, then really, what does it have?

My one problem, as I've said repeatedly, is that there are certain very important factors that no numerical model can possibly take into account. (I did also have a problem with Aaron's decision last season to put the Colts at number 1 with an asterik, but that's neither here nor there.)

 

[thatguy]

But team A's coach gets conservative and resorts to the run. Team B is expecting the run, so they sell out to stop it. Team A has a lot of stalled drives, which is fine because they're sitting on a lead. Because most of these runs will likely be worse than the league average due to Team B's adjustments, these drives will net Team A a lot of negative DVOA.

Not true. Team A's run will be compared to all other teams who run the ball with an early 21 point lead, not to all runs, period. So if there's a negative hit there, it'll probably be pretty small.

Ah, thanks for clearing that up. I suppose then the one thing it still won't take into account is coaching philosophy, but that makes me much happier.

Abrecher, how precise does DVOA get when determining game situation? By this, I mean, you say they'll compare it to all other teams who run the ball with an early 21 point lead. What does this mean, exactly? What is early? And will they also compare it to teams with a 17 point lead? A 24 point lead? Etc.? I think you get what I'm asking. If not, here's an example: I doubt they get as precise as something like comparing the results of all teams who run the ball at the opponents 44 yard line on 2nd and 8 with 8:35 left in the second quarter when leading 21-0. So, what are the perameters used to determine a "similar game situation"?

 

[Sea Bass]

I don't want to sound like a homer, but the Seahawks at #23...wtf?

I know we have had injuries and lost Hutch in the offseason. This is plain crazy talk!