2009 Preseason DVOA Projections
by Aaron Schatz
OK, folks, here they are: the 2009 DVOA projections.
We must start with the requisite link to an explanation of DVOA. For anyone new to our site, DVOA stands for Defense-adjusted Value Over Average and measures a team's performance on every play of the season compared to league average in the same situation, adjusted for opponent. I know a lot of people are coming here from various message boards and this is just going to look like a jumble of pointless numbers. Trust me, there is a method to the madness, and over the past eight seasons DVOA has been a far more accurate predictor of future performance than wins or points.
Offense, defense, and special teams DVOA are all projected separately using a system based on 2000-2008 numbers. The equations include a number of variables based on performance over the past two seasons in different splits (by down, passing vs. rushing, red zone vs. whole field) plus variables based on recent draft history, injury history, offensive and defensive pace, coaching experience, quarterback experience, and even weather. Strength of schedule was then figured based on the average projected total DVOA of all 16 opponents for 2009 (yes, projected performance, not 2008 performance).
The projections here are updated from Football Outsiders Almanac 2009 based on changes in some of the variables, usually related to injuries, offensive line continuity, and quarterback experience. There are no manual adjustments. The numbers we are presenting here are exactly what the projection system spit out. As we say every year: "A few of them will look strange to you. A few of them look strange to us." There's no reason to really regurgitate the reasons why our projections for some teams (Jacksonville, Green Bay, the entire NFC West) differ so greatly from the conventional wisdom -- we've gone over it in numerous articles and in numerous promotional interviews over the past few weeks. We know that there are a couple of teams whose projections in offense and defense are far beyond the rest of the league. That's the system expressing a high amount of confidence.
This year there has been very little change in the projections during the preseason. Some notable changes include:
- San Diego dropping from "completely absurd" to a more reasonable "best team in football."
- Seattle dropping a bit because of injuries.
- Detroit moving up because of variables related to the average age of starters in various units.
- Minnesota moving up, a little bit because of Favre but more because we no longer are penalizing them for only having the Williams Wall for 75 percent of the season.
- Carolina's defense improving now that Julius Peppers is securely in the fold.
The projected wins are going to be a little different from those you found in FOA 09. Rather than being based on Ben Alamar's simulations, these are based on the simulations we run to figure the DVOA Postseason Odds Report each week. There are two differences between the two simulations:
1) The FOA 09 simulation gives each team a different DVOA rating in each run, based on a normal distribution in all the different variables we use to project DVOA. The playoff odds simulation, on the other hand, only uses mean projected DVOA -- the quality of each team stays the same in each run, and the only thing that changes is which team wins each game.
2) The way the playoff odds equations figure the chances of winning each game end up dragging every team even closer to the mean of 8-8.
Projected division champions are colored in light yellow. Projected wild card teams are colored in light blue. At first glance, it may look like we're predicting that the AFC will regain its total dominance over the NFC, but if you look a little closer you'll see that isn't the case. The AFC is very stratified compared to the NFC. Thirteen NFC teams are projected between seven and ten mean wins, compared to just five AFC teams. Our projections draw a clear line between the top seven AFC teams and the rest of the conference, but the NFC, particularly the NFC South, is more of a free-for-all.
The first full playoff odds report can be found here. Tomorrow, we'll have our annual look at subjective staff predictions for 2009.
160 comments, Last at 13 Sep 2009, 1:06am
#1 by oakfan (not verified) // Sep 09, 2009 - 6:39pm
Looks great! Thanks for posting!
After a cursory look at these, the most surprising thing to me was Green Bay's defense ranked 24th after finishing 12th last year. Is a lot of this simply due to switching to the 3-4?
#41 by Arkaein // Sep 09, 2009 - 9:43pm
That is a bit baffling. Even after looking over the GB FOA09 chapter again I can't find any really major indicators for defensive regression.
Change in scheme on average produces negligible change in DVOA. The Packers had absurd scoring from their defense early last year, but this boosts Pythagorean projections, DVOA already assumes that pick-sixes are mostly random. Their 3rd down defense was worse than first or second down, which should indicate a slight improvement (on the other hand the offense is expected to decline in large part to absurdly good 3rd down offense compared to earlier downs). The packers also didn't have any significant defensive personnel losses, while adding two first round draft picks.
The only things I can think of as downward trends are the age of the CBs (expecting Harris and or Woodson to decline or miss significant numbers of games due to injury), and the general downward trend of the defense last year from pretty good in the first half to pretty bad down the stretch.
Considering that the late season defense was missing Cullen Jenkins and Nick Barnett, with Al Harris missing about four games towards the middle, I would think that health trends from 2008 to 2009 would be at worst neutral though.
#2 by jonnyblazin // Sep 09, 2009 - 6:43pm
Sorry, if the Ravens have that bad an offense I'll eat my hat. Do the projections factor in how crappy a coach Billick was, and how good Cameron is? Flacco has been really good this preseason, and its already established that the Ravens have a superior offensive line that is still young and improving.
#3 by jonnyblazin // Sep 09, 2009 - 6:49pm
And what about the Steelers having an average D? Zany. I love these projections by the way, but the Ravens and Steelers #s look strange.
#4 by Will Allen (not verified) // Sep 09, 2009 - 6:52pm
Well, I do think Capers will achieve more with the switch of scheme in Green Bay than Aaron thinks is likely, but I wouldn't bet a bottle of 15 year old Laphroiag on it. Otherwise, I pretty much agree with his assessment of the NFC North, and think the division will be decided by injury luck, and not just with 40 year old jean models.
Actually, injuries, guys shooting themselves in nightclubs, guys getting arrested for allegedly assaulting reality t.v. bimbos named afte liqour, etc., will likely play a huge role in determining all division champs, as much as analytical soothsayers like us try to believe we can rationally peer into the future.
#6 by Aaron Schatz // Sep 09, 2009 - 6:57pm
I'm only going to break in once to remind people: This is not "Aaron thinks." This is "the numbers think." I don't think the Steelers defense will be average, and I don't think the Jaguars defense will be this good. These are statistical projections based on a formula -- the best formula, we believe, but not a perfect one, and not one that takes into account my subjective opinions.
These numbers also represent the average of a range of possibilities. All the subjective issues that we know play a role in wins and losses -- teams being healthier or unhealthier than average, young units suddenly gelling with experience, Mark Sanchez being the second coming of Matt Ryan, players attacking their MTV hostess girlfriends, etc. -- create that range of highs and lows.
#12 by greybeard // Sep 09, 2009 - 7:13pm
Aaron's opinions played no role on the formula therefore this is not "what Aaron thinks". God gave the formula to Aaron, so it is what God thinks. The formula has no subjective opinions in it because it was written by God. Aaron is the football prophet of God. He is just the messenger.
#17 by Jerry // Sep 09, 2009 - 7:38pm
If you have a better formula, as opposed to what you think, by all means let us know what it says.
#27 by greybeard // Sep 09, 2009 - 8:25pm
I don't think you understand it. I cannot come up with a formula myself because that would be subjective, what I put in it would be entirely based on my personal opinions. Whereas DVOA does not suffer from this because it is based on laws of nature, created as God intended.
#30 by Never Surrender (not verified) // Sep 09, 2009 - 8:50pm
LOL. I like to take my shots at the many stat fans who, in my opinion, blur correlation / causation issues and read way too much predictive power into past performance. So I find the occasional joke like this to be hilarious.
That said, though, I really appreciate the efforts that are going into DVOA. It's a very interesting metric, and not in a trivial way. Definitely something worth doing and discussing.
#66 by D Jones // Sep 09, 2009 - 11:55pm
True that. DVOA is a model, and much like the quant guys at Goldman Sachs used to say, "there ain't no perfect model." Okay, they didn't say that, but they should have.
#68 by Independent George // Sep 10, 2009 - 12:37am
Are you then saying that if the model doesn't match what his personal opinion is, then he should tweak the model until it does, thereby making it less subjective?
#110 by greybeard // Sep 10, 2009 - 1:02pm
Not really. I am saying that he should just admit that the model is a reflection of his and his staff's opinions. Opinions that are not about the teams or their particular standings, but based on what his and his co-workers perception about what constitutes success on first downs, seconds downs, third downs, etc and what constitutes as failure on interception, etc. Is an interception worth one punishment points, two, five?
One can claim that getting 3 yards on a first down is success but DVAO may think that it is a failure. That is the subjective part.
Besides no-one knows how DVOA is calculated. It is not public, it is not peer reviewed, it is not reproducible by independent third parties, etc. So if they have a bug in their software that calculates the DVOA, we will never know that, neither will they. If it is entirely bullshit we would not know. Maybe it is based on random numbers on a Gaussian curve where the median is moved based on the opponents conventional statistics corrected for strength of schedule. That would be an opponent adjusted stat, but yet would be entirely bullshit. I am not suggesting that DVOA is bullshit. I am pretty sure it is nicely crafted, well thought out metric. Yet without knowing what it really is, I will refrain from seeing it as something objective.
To suggest that everything is based on formula and therefore must be objective is not correct.
I love this site. There are many great articles on the X and Os, drafts, team analysis. The people who come here and discuss football are mostly intelligent and civil.
But I believe the statistics aspect of the site is really not something I care much about. The "forecast" part of the Lewin projection was entirely bullshit in my opinion. Yet I saw that anytime draft and QBs were mentioned, people would use that as a means of proving their point. I think a stat like DVOA is sensible for teams. But for individual players, for QBs, WRs, it is really not useable. There is soo much that affects the success of QB beyond his abilities and performance, to suggest that it can be measured from game to game, or within a season is just bad science. Maybe for the entire career of the player DVOA makes sense.
It is up to the reader of this site or the book to decide how valuable DVOA is or not. I have my opinions about the value of the DVOA and so do others. But regardless of the value one assigns to it, it is by no means objective. That was my point.
#114 by steelberger (not verified) // Sep 10, 2009 - 2:29pm
My question is: what formula is it that predicts the Steelers to have barely adequate defense (13th was it?) despite them having the best defense in the league for two years straight (by yardage), being a top 10 defense for 9 years straight, and having a much softer schedule than they had last year.
#116 by Thomas_beardown // Sep 10, 2009 - 2:45pm
DVOA is supposed to account for opponent, so a softer schedule shouldn't lead to a better DVOA ranking.
#125 by Eddo // Sep 10, 2009 - 3:45pm
I don't think you really understood greybeard's point at all; he wasn't pointing out that "because Team X's projection looks weird, the whole system is wrong" like you are. Rather, he was pointing out that he's skeptical of a system that is not open-source.
He has a point, but at the same time, Aaron is under no obligation to reveal the formulas behind DVOA. His goal is to provide accurate analysis and make money, not provide accurate analysis and get published in a scientific journal.
#137 by steelberger (not verified) // Sep 11, 2009 - 7:11am
I understood what he said perfectly. Formulas like these can (and are) tweeked until they produce the results that the "creators" feel is correct. That cannot possibly be said to be non-subjective.
My question about the Steelers was separate from that and was just an example of how the formula itself is flawed.
#138 by Arkaein // Sep 11, 2009 - 8:56am
Actually, the results are tweaked to produce the most accurate predictor of future wins. Saying that they are tweaked to produce results Aaron and co. "feel is correct" ignores the fact that all of the factors used in DVOA are tweaked by a computer using regression analysis.
People need to select the factors that are analyzed, but the algorithm determines how important each factor is.
#142 by Chip // Sep 11, 2009 - 11:57am
This quote (from the 2009 staff predictions article - who will outperform their FO projection question) makes greybeard's and steelberger's point exactly
"Aaron Schatz: Arizona. In interviews, I've said that I spent the spring constantly looking for ways to get the Arizona projection higher, and I meant it."
The interview he referenced was with Arizona media / blog and he made further incriminating comments about changing variables and such (don't have the link)
Come on, the ENTIRE projection system is subjective.
#145 by Eddo // Sep 11, 2009 - 12:47pm
I've been interpreting those comments differently that you, I guess. Here's how it was playing out in my head:
Program spits out projections.
Aaron: "Gee, Arizona's really low. Let me lessen the impact of having an old QB."
Program spits out projections.
Aaron: "Hmm. Now more teams have gotten less accurate. I guess I can't change the having-an-old-QB weighting. Maybe I can tweak the having-a-single-great-receiver weighting."
Program spits out projections.
Aaron: "That also made things less accurate on the whole. I guess I'll try..."
Goes on for a while...
Aaron: "Well, it looks like I'll just have to leave Arizona lower than I personally would have ranked them."
I mean, you're right that the rankings are subjective elements, but it's not like Aaron only tweaks the numbers for certain teams. He has definitely said in the past that, when he alters some factors to make one team come out higher or lower, he applies the same changes to every team, and if DVOA gets less accurate, he scraps those changes.
#146 by Arkaein // Sep 11, 2009 - 1:01pm
It seems that Aaron tried to make Arizona do better in the projections because he felt the projections were flawed in how they handle the playoff surges that have occurred not only for Arizona, but for other teams in recent years.
However Arizona's projection is still bad. Why? Obviously The factors that Aaron tried didn't improve DVOA's predictive power. That's why the ENTIRE system is not subjective, because the somewhat subjectively selected input factors need to show predictive capabilities to become integrated into the system.
DVOA isn't perfect. However it has improved steadily by ensuring that changes improve predictive power. That seems fairly objective to me.
#22 by Temo // Sep 09, 2009 - 8:08pm
I have to admit, that was pretty funny.
#33 by Will Allen (not verified) // Sep 09, 2009 - 9:04pm
Yeah, I'm laughing too. I have a vision of Aaron (long beard version) as Mel Brooks (History of the World, Part 1), as Charleston Heston, playing Moses, three tablets in hand, saying "I have here the Fifteen," (drops one tablet, shattering) "er....Ten Commandments of Football Analysis", which explains how Joe Theismann ends up in the broadcast booth...
#59 by D Jones // Sep 09, 2009 - 11:34pm
I think those tablets also explain that whole Matt Mullen thing.
I was also digging the Aaron has a direct link to God thing, not as good as Moses' link, but still, not too bad.
#5 by rk (not verified) // Sep 09, 2009 - 6:54pm
The Steelers winning with offense. I suddenly wish I had bought FOA to read about how that's going to happen.
#7 by JasonK // Sep 09, 2009 - 6:59pm
The Jets offense and the Dolphins defense being projected as dead last are conversation starters, to say the least...
#10 by Telamon // Sep 09, 2009 - 7:12pm
Since when is saying a rookie QB with no receivers will be bad going out on a limb?
#92 by silentdibs (not verified) // Sep 10, 2009 - 9:38am
Provided the line stays healthy, I think the running game is sound. I could see 20th, 25th place, but dead last? And it's not even that close, except for Buffalo. Heh heh, Buffalo.
#13 by DrewTS (not verified) // Sep 09, 2009 - 7:19pm
I don't know about anyone else, but I'm not really looking for big things from the Jets offense this year. Maybe not dead last, but not in the top half of the league.
Matt Ryan has ruined the expectations for rookie QBs for the foreseeable future.
#8 by Insancipitory // Sep 09, 2009 - 7:06pm
Some of that stuff looks crazy, but remember the 2005 Buccs.
#61 by Scott C // Sep 09, 2009 - 11:36pm
and the 2004 chargers, and other "no freaking way" projections that weren't just dumb luck: they projected things for reasons that turned out to be what happened.
I don't think Pittsburgh's defense is going to decline like this projection, but if it does I'd be interested in looking into whether the reasons for the decline in reality are similar to the projection's reasons.
#9 by DrewTS (not verified) // Sep 09, 2009 - 7:09pm
Kudos to someone (even if it is just a computer model) having the guts to predict that Arizona will completely tank this year. I've been wondering for a while why everyone seems to be penciling them in for a cakewalk back to the playoffs. That division was TERRIBLE last year, and they went a whopping 9-7 in it. One good playoff run does not a juggernaut make. And if you believe in the Super Bowl Loser's Curse, go ahead and throw that one in there too.
#98 by Eorr (not verified) // Sep 10, 2009 - 10:49am
Aaron I think made it pretty clear on the Sports Guy's podcast that the Arizona regression has alot to do with Warner's age and how the formula penalizes a person for being a 38 year old quaterback who was unusually healthy for all of last season.
#11 by bravehoptoad // Sep 09, 2009 - 7:13pm
I'm gonna go cry in my cups. Poor '9ers.
#18 by beargoggles // Sep 09, 2009 - 7:38pm
Well, the local media is drinking the Kool-Aid, despite poor QB situation and no passrush. Sounds like the run game may be excellent, but poor 9ers again. :(
#14 by slipknottin // Sep 09, 2009 - 7:21pm
chicago with the 2nd best defense? Hmm. I have a lot of trouble believing none of the NFC east teams will finish in the top 10 in defense (ok so the giants are ranked 10th) All 4 of them were 10 top last season (ok, so redskins were 12th) Still.
#15 by Insancipitory // Sep 09, 2009 - 7:25pm
Haha it predicts Seattle via Denver will have the number 1 overall pick. I sort of want that to be true, and hell let's just throw that in with a longshot superbowl win.
#16 by Joseph // Sep 09, 2009 - 7:36pm
Do these rankings take into account the firings of OC's Chan Gailey, Turk S., and Jeff Jagz? Cause if not, BUF is REALLY gonna stink, since they waived LT Langston Walker.
#19 by mm (not verified) // Sep 09, 2009 - 7:42pm
I was surprised that it already projects 2 teams to have more than a 90% chance to make the playoffs, but then I remembered which divisions those teams played in.
Probably the most interesting projection is the top 2 teams in the NFC being from the North, although I suppose the suspensions being reinstated could bounce Minnesota down again.
Did you remove the Saints 2 suspensions for this projection as well? To the surprise of everyone covering the Saints this week, they didn't start their suspensions; the NFL is apparently linking them to the Vikings' case, even though that is about Minnesota state law. Unfortunate, if they eventually do have to serve the suspensions, because this is the easiest part of their schedule.
#21 by Insancipitory // Sep 09, 2009 - 8:00pm
I wonder what the fallout from that is going to be. Would the Williams(es?) get off free and clear? Would everyone get a pass? Would they be suspended for away games? Should I make a sign for the Seahawks games with a drawing of the state that says, "No Income Tax. You're Welcome."
#58 by BigCheese // Sep 09, 2009 - 11:29pm
So. why is SD missing from the Top draft pick predictions? Is it because botht heir numbers show up at 0%?
#20 by Rover (not verified) // Sep 09, 2009 - 7:45pm
The past 2 years I took the DVOA Aaron's mean win and compared them to Vegas O/u lines. I bet the top 6 disparities (weighting my bet by the amount of difference). Both years I made money, although last year was very close.
If you can beat the odds makers, I'd say you ahve an effectvie engine. I'll let you know about next year.
#50 by Big Johnson (not verified) // Sep 09, 2009 - 10:33pm
nice rover. If i wasn't living off the poor mans cup of noodles every night i would copy your idea. Ive been wondering for a couple of months if that would work. Thanks for the post and let us know next year how it goes
#23 by Soulless Merch… // Sep 09, 2009 - 8:17pm
So...who will Buffalo be picking #1 overall next year?
#88 by Mike B. In Va // Sep 10, 2009 - 8:23am
Sadly, there's no Bruce Smith available like there was last time this happened.
#121 by Bobman // Sep 10, 2009 - 3:25pm
Can you draft an O Coordinator? The great thing is they'd only have to pay him from draft day til Labor Day, when he gets fired.
#24 by afootballfan // Sep 09, 2009 - 8:18pm
Steelers defense 13th? What, is this where FO predict a a mass amount of injuries
to suddenly take place? lololol
I see Patriots up near the top. Classic FO.
#25 by Geo B // Sep 09, 2009 - 8:22pm
Agreed - no way Steeler defense drops to 13th unless two significant season-ending injuries happen - Football Gods forbid! Even as a Steeler fan, the DVOA computer blew it on that one. Maybe it's not weighting recent performance enough or something. I'll also take the OVER on 9.9 wins ;-)
Steeler fan trapped in Houston!
Six Time SB Champs! ;-)
#38 by Jerry F. (not verified) // Sep 09, 2009 - 9:30pm
That is classic. Year after year the Patriots suck, but for some reason FO keeps predicting they'll finally turn things around and be one of the best teams in football. It's like Berman's 49ers-Bills schtick, but less plausible.
#47 by mattymatty // Sep 09, 2009 - 10:21pm
The Patriots went 11-5 last year which is hardly sucking.
#64 by richardbrianwa… // Sep 09, 2009 - 11:49pm
When was the last time the Patriots sucked?
#76 by Brendan Scolari // Sep 10, 2009 - 4:51am
Wow, the two posters above fail at detecting sarcasm.
#26 by drobviousso // Sep 09, 2009 - 8:24pm
I would love to hear how Pittsburgh develops a top 5 offense and an average D.
#48 by Israel P. (not verified) // Sep 09, 2009 - 10:22pm
Maybe system remembers that Marvel Smith and Kendall Simmons were injured last year and expects them back.
#112 by Matt W (not verified) // Sep 10, 2009 - 2:20pm
Smith retired and Simmons signed with New England last weekend. Maybe the system looks for improvement from just having the same five guys together longer? Still, seems odd.
#28 by Led // Sep 09, 2009 - 8:34pm
My subjective impression is that DVOA has the Pats' offense too low and the defense too high. Same for Chicago. They both probably balance out in total DVOA. Seattle being that high in offense is also a suprise. And although I understand the reason why DVOA is down on Baltimore's offense, I think Flacco is going to be a good player.
I'll admit this could be homerism, but I also would be very surprised if the Jets are dead last in offense, especially in any league that still has the Bills. I fully expect mediocre to bad, but outright, cover your eyes terrible would surprise me. Even with a rookie QB and an a no name collection of receivers, the OL, backs and starting TE should keep them from being embarrassing. But we shall see.
#31 by Mr Shush // Sep 09, 2009 - 8:51pm
Agreed. No team with an OL as good as the Jets' could ever be the worst offense in the league. I'd say Buffalo and Detroit were the most likely contenders for that Crown, while St. Louis and SF are both also in with a shout. Like everyone else, I cannot conceive of a way in which Pittsburgh could have a top five offense this season. Their line is terrible. Their backs are terrible. Their quarterback and receivers are good, but not spectacularly so.
#143 by TGT2 (not verified) // Sep 11, 2009 - 12:17pm
The more than 50% turnover on OLine, including a 57 year old Center and a Rookie RT probably are the main reason.
Flacco was average last year, and behind an expected pourous line with his top wideout now 73, I doubt the system expects great things from the 2nd year QB. Those indicators point to suffering in the run game, too.
If everyone stays healthy and the OLine and Flacco continue to perform like they did in the preseason*, then the projection will be low. If Mason and the new Center get injured, and the OLine acts like a newly formed OLine, the projection might even be high.
*I do not trust the preseason. If I did, I'd be saying that Flacco is a 2nd Ballot Hall of Famer. Despite that, I'm cautiously optimistic, as any good non-philly fan is this time of year.
#29 by jpo287 // Sep 09, 2009 - 8:49pm
I know better than to argue with the numbers by now but still, Carolina's numbers are flipped from last year. Given they have the same offensive line, may have found a slot receiver in Moore, still have Williams and Stewart and may have found a gem in Goodson, I just don't get it. Is Jake just going to fall apart this year? I know there's not a bunch of love for him around here but still!
And defensively, they lost Keemo, who while not great did take up a lot of space, and yet they are ranked 3rd?? I'll take it but really?
I love how you have Atlanta and while I never expected them to repeat last I never would have expected them to fall so far!
I know football season is finally here with my weekly stats reading/rankings. THANKS!
#32 by Never Surrender (not verified) // Sep 09, 2009 - 8:55pm
Steelers fans crack me up.
#34 by morganja // Sep 09, 2009 - 9:04pm
I'm a bit confused as to why the NFC South has fallen to the worst division in Football after being the best last year in wins. The Panthers winning the division at 8.6 - 7.4? Really? Last year they were without a losing team at 12-4, 11-5, 9-7 and 8-8. And FO seems to think they have an easier schedule? I don't have any resaon to doubt what would be the most spectacular fall of a division in history, but there doesn't really seem to be any plausible reason for it. Is it possible that there is a systemic error somewhere? Is there any hint as to why the NFC South will suck so bad this year?
#44 by JasonK // Sep 09, 2009 - 9:52pm
#94 by Dr. Jacques Therape (not verified) // Sep 10, 2009 - 9:58am
The NFC South as a whole has much a much harder schedule this year. Last year, they played the AFC West & NFC North -- this year it's the NFC & AFC East. Last year, for example, the Bucs had probably the most favorable schedule in their history, with no west coast trips and no cold weather games.
#102 by morganja // Sep 10, 2009 - 11:34am
That would be my subjective opinion as well as far as win-loss goes. But FO actually claims that the schedule is easier this year. Note the strength of schedule for the NFC South: 25th, 29nth, 30th and 31st. Their DVOA is also just horrible irregardless of the competition, -.5%, -8.1%, -11.9% and -17.2%.
It just deosn't make any objective sense that I can see. If we were talking about the AFC West with Denver and the Raiders having complete off-season meltdowns I could see it. But what has happened to cause this dramatic collapse for the entire division?
It honestly looks like there is something wrong with this years projections, as in a misplaced modifier or a divisor error in the math. I've been following FO for a long time and have been mildly surprised before, but thought their predictions plausible. This is the first time I have absolutely no faith in their predictions.
#107 by JasonK // Sep 10, 2009 - 12:31pm
At the risk of getting circular here, the NFCS teams' strength of schedule is weaker because the NFCS teams are projected to be weaker than they were in 2008. Take CAR, TB, NO, and ATL out of the top half of the DVOA rankings, and that offsets the increase in non-divisional difficulty.
Among the factors upon which I think the system is basing this projection:
- a return to average injury rates in Atlanta and Carolina
- age-related decline for Jake Delhomme, Steve Smith, Tony Gonzalez, and others
- "new coordinator" adjustments for TB, CAR defense, and the NO defense
- Byron "Trebuchet" Leftwich starting at QB for the Bucs
#115 by morganja // Sep 10, 2009 - 2:37pm
Does it count as a new coordinator in NO? I thought they just hired Williams to replace "Hole in Coaching".
#35 by Dales // Sep 09, 2009 - 9:15pm
I think the formula would end up being more accurate if the spell checker swapped GB with Chi, in this instance.
#84 by Jimmy // Sep 10, 2009 - 7:51am
That would give the Pack the 20th ranked offense and the 2nd ranked defense and special teams. It isn't all that good a fit.
#91 by Dales // Sep 10, 2009 - 9:33am
I wasn't going after similarity scores. I was going after 'where I subjectively believe the two teams will end up.'
I think Cutler is a very good quarterback. I do not think he is good enough to make the Bears one of the top 8 teams in the NFC, no less one of the top 4 in the NFL and best in the NFC.
Granted, I don't think GB is that good either, but I believe they are closer to it. And, yes, I also know that Cutler is not the only reason the Bears rate so high in the system. I just am taking it with a grain or five of salt-- and just take it that there are reasons for me to think that I was undervaluing the Bears, even if I don't accept how much.
#132 by Charlie (not verified) // Sep 10, 2009 - 5:47pm
I know 2008 win-loss isn't everything and they could easily get worse this year, but they were 9-7 last year; you could argue they were a top-8 NFC team before they traded for Cutler.
Having said that, as a Bears fan, it does feel like DVOA is being too generous to us this year. The #2 Defense, the best team in the NFC? Shtop it, you are shpoiling us.
#36 by Temo // Sep 09, 2009 - 9:20pm
The correlation between 2008 Predictions and 2008 actual DVOA was 0.481. The absolute value of the difference between predicted and actual DVOA in 2007 had a mean of 16.7% with a standard deviation of 11.7%. The 2008 predictions missed most on Atlanta (underestimated by 50.1%) and Seattle (overestimated by 42.3%) and missed least on Philly (off by only 0.6%), Arizona (off by 1.7%) and Buffalo (off by 2.0%).
Interestingly enough, over the 3 years I have this data, the predictions have missed the most by far on Seattle (off by a combined difference of 107.2%), mostly because of overestimating the team by tons in 2006 and 2008. Arizona has the best-guess team thus far, with a 3-year combined difference of just 10.8%-- predictions for Arizona were the among the most accurate in the league in both 2006 and 2008.
The correlation between 2007 Predictions and 2007 actual DVOA was 0.582. The absolute value of the difference between predicted and actual DVOA in 2007 had a mean of 13.6% with a standard deviation of 9.7%.
The correlation between 2006 Predictions and 2006 actual DVOA was 0.358. The absolute value of the difference between predicted and actual DVOA in 2006 had a mean of 15.8% with a standard deviation of 11.5%.
#39 by Temo // Sep 09, 2009 - 9:30pm
Also interesting to not that in last years' thread, people seemed to think the strangest picks were Philly (too high), Dallas (too low), and TB (too high). DVOA predictions on all 3 of those turned out to be very accurate. Its the two that no one had too much problem with (Seattle and Atlanta) that threw off the predictions.
#80 by bubqr // Sep 10, 2009 - 6:13am
Thanks an awful lot. Great information.
The NFC East Defenses predictions is quite shocking to me (PHI 18th, is that all Bradley + Dawkins ?), but also are ATL and GB odds of making the playoffs : 16 %. Wow. Plus STL that high in DVOA too ?
Phi offense not much liked by DVOA : I'll gladly tak the over on this one. OL stability is important, they never had as much talent.
#85 by Jimmy // Sep 10, 2009 - 7:56am
Bradley, Dawkins and Jimmy Johnson perhaps. That could hurt a team.
#67 by D Jones // Sep 10, 2009 - 12:05am
The correlation analysis is quite interesting. Thanks.
I was wondering just that while I was looking at the tables...
#83 by daveb (not verified) // Sep 10, 2009 - 6:24am
Those are pretty low correlations...and high absolute errors. I bet most readers of this site could get close to that predicting total DVOA to that degree of accuracy. In fact that would be a fun thread.
I think DVOA's value is really its retrodictive ability.
#101 by Temo // Sep 10, 2009 - 11:28am
2006 was especially bad, but the past couple of years have been pretty decent. It's nothing equaling the accuracy of, say, BP's baseball predictions, but it's better than the majority of lay people can do. I doubt the average NFL prognosticator could beat the statistical evaluations of teams' chances, though I suppose very knowledgeable insiders probably could.
#37 by Fan in Exile // Sep 09, 2009 - 9:24pm
How have their predictions done for Denver, because right now it's looking pretty absurd.
#40 by Temo // Sep 09, 2009 - 9:35pm
Denver is in the middle of the pack. Other than Seattle, where the DVOA predictions missed by huge amounts for two seasons, no other team has been misjudged significantly in more than one season.
They got Denver right to within 12.1% DVOA last year, 3.0% DVOA the year before, and 22.0% the year before that. With a Standard Deviation of about 15% DVOA, really only in 2006 did they miss Denver by any kind of noticeable amount.
#42 by Key19 // Sep 09, 2009 - 9:45pm
#49 by randomviking // Sep 09, 2009 - 10:30pm
Note from a non-Broncos fan who lives in Denver: the media here don't really seem to be preparing people for what a train wreck this season is likely to become. I really don't think that your average Broncos fan thinks that the team is as bad as it's likely to be....
#51 by Insancipitory // Sep 09, 2009 - 10:40pm
Good, their suffering will only be sweeter.
#120 by Piglet (not verified) // Sep 10, 2009 - 3:24pm
I'm a Denver fan, and have been greatly impressed by the (admittedly imperfect) DVOA predictions from year to year, but my subjective assessment is that Denver is actually pretty unlikely to be the second-worst team in the league this year. They weren't the second-worst team in the league last year, they look to me like they are better at almost every position except for QB, and I think that Orton will prove to be about a league average, not bottom-of-the-barrel, QB. My hypothesis is that DVOA is likely to be less predictive when it comes to teams that have undergone such significant changes over an offseason, and that by the end of the year, Denver will look a lot more like a middle-of-the-pack team than a historically bad team using DVOA. (The Broncos record may be bad due to what looks like it will be a brutal schedule, but I fully expect them to be competitive most weeks.)
No trash-talking here, there is ample reason for people to be skeptical of the Broncos, and the great thing about DVOA is that, looking backward, it is an objective (if imperfect) assessment. If the Broncos DVOA actually ends up at -30% I'll freely admit I was wrong, but as of right now I don't think that his how it is going to go.
#123 by Thomas_beardown // Sep 10, 2009 - 3:38pm
I agree with you about Orton, he is close to average. However, how exactly did the Bronco's receivers and o-line get better?
Also, most of the defense could improve substantially and still be among the worst in the league.
#129 by Piglet (not verified) // Sep 10, 2009 - 4:29pm
Eddie Royal with more experience -- he looks great. Brandon Marshall had his hip fixed (assuming he gets his head on straight, open issue there). I liked the Gaffney pickup since he knows the new system. Not much improvement on the O-line, I'll admit, but there wasn't that much improving to be done there. D is a huge question mark, obviously.
As I said, lots of questions and reasons to doubt. We'll just have to see what we see.
#87 by Mr Shush // Sep 10, 2009 - 8:05am
I'm afraid to say that until recently my subjective expectation was that Denver would be the worst team in the league. The Schonert firing and accompanying hoopla has persuaded me that Buffalo will be worse. On this point, my instincts and DVOA's projections appear to be in total agreement.
Take the second worst defense of the last decade and a half. Add another year of ageing for the already prehistoric secondary. Remove franchise quarterback and pro-bowl wide receiver. Change little used blocking scheme with unusual talent requirements which has enabled line to be the best in the league to one for which there is no guarantee the personnel are suited. Sounds like a recipe for disaster to me.
#43 by maxpower179 // Sep 09, 2009 - 9:48pm
DVOA has a history of predicting performance more accurately than any other formula I know of - which is not necessarily to say more accurately than a sharp human being, and it's definitely not perfect. It wasn't given to Aaron by God, it was developed over the years through a lot of sweat and tears. I'm shocked by a lot of the numbers too. But that's what makes it interesting. This is my second favorite part of the beginning of football season, next to the opening kickoff itself.
#45 by Xeynon (not verified) // Sep 09, 2009 - 9:54pm
Kudos to someone (even if it is just a computer model) having the guts to predict that Arizona will completely tank this year. I've been wondering for a while why everyone seems to be penciling them in for a cakewalk back to the playoffs. That division was TERRIBLE last year, and they went a whopping 9-7 in it.
Is DVOA's prediction really that much of an outlier? Although I can't recall specific names, it seems to me that I've read quite a few other, more conventional prognosticators who have also predicted them to stumble this season. Certainly there are lots of factors (Super Bowl hangover, the Seahawks being unlikely to lose half their team to season-ending injuries again, Kurt Warner being unlikely to make it through sixteen games healthy at his age again, etc.) to do so.
#46 by tally // Sep 09, 2009 - 10:18pm
Looks like the Pats have the easiest schedule again.
#52 by ernie cohen (not verified) // Sep 09, 2009 - 10:41pm
Aaron, the rules of science say that you can say "this is just what the formula says" only if you tell us what the formula is. Without us being able to look at the formula to check that it's sensible, or rerun the numbers for ourselves, why in the world would we believe these predictions? Would you?
Our problem has nothing to do with conventional wisdom. It has to do with the fact that the PIT D 2000-2008 DVOA numbers read
-14.3, -10, -4.5, -3.3, -18.5, -15.6, -7.0, -11.8, -26.9
The "formula" now says -2.0. As far as I can tell, none of the factors mentioned in the introduction hint at why the formula predicts the sudden collapse of the PIT D. Is it based on something else, like age? Or should we suspect something more mundane, like a programming error?
If anybody really does believe this prediction, I'll happily bet up to $20K that the PIT D DVOA ends the season better than -2.0. I'll be nice and give you -4.0. Any takers?
#108 by Ben Stuplisberger // Sep 10, 2009 - 12:36pm
It's not scientific if its proprietary.
#151 by Jerry // Sep 11, 2009 - 4:33pm
So no pharmaceutical company research is scientific?
#153 by Ben Stuplisberger // Sep 11, 2009 - 5:51pm
Don't they have to reveal how they reach their conclusions? I don't know, that's why I am asking.
Pharmaceutical research can be scientific, but often, it is not.
#126 by Eddo // Sep 10, 2009 - 3:46pm
So going from -3.3 to -18.5 is perfectly reasonable, but going from -26.5 to -2.0 is so questionable that you'd be willing to put $20,000 on it?
#147 by erniecohen // Sep 11, 2009 - 2:29pm
No, I would have been pretty happy betting against it going from -3.3 to -18.5 too (though not nearly as happy as I am to bet it's not going down to -2.0 this year). Remember, the number that comes our of the prediction is the mean (or perhaps median) of a number of trials. I have no problem with the idea that -3.3 was worse than expected, or that -18.5 was better than expected.
My point was that it is said up front that the predictions are based on DVOA data from the previous years, plus some other enumerated factors. I'm just trying to understand where this prediction is coming from, because it sure as hell is hard to reconcile with the numbers from the last 8 years.
Remember, DVOA is supposed to be (1) to be less noisy than ordinary stats and (2) to convey *understanding* of why teams win and lose. There is some purported understanding that the formula seems to be claiming here, and I just don't see what it is.
#148 by Eddo // Sep 11, 2009 - 2:44pm
I think you're correct, but I'm leaning more towards last year's -26.5% defensive DVOA being a bit of an outlier*, so a mean of -3.3% is all that shocking. Granted, it still surprises me that it's a little low, but not a reason to trash the entire set of projections.
* What I mean is that the Steelers' defense was probably not a "true talent" -26.5% last year; rather, they were probably in the -12% to -18% range. It'd be like having a six-sided die with four W's and two L's. It would be a "true" 66.7%-W die, but over stretches you'd sometimes get 80% or 90% W's.
#149 by erniecohen // Sep 11, 2009 - 3:42pm
I would also guess last year is a bit of an outlier, primarily because they had somewhat better than average injury luck on D (given their ages and injury history). And maybe they really are a -12-16 range D (though I don't see much reason to believe so). But that just supports my point: why would you expect the "true" level of the D to drop from -16 to -2 in a single year?
Again, I'm not trashing the predictions; I'm trashing the methodology. I'm frustrated that I can't understand what the hell is going on in these predictions, and I can't find out because we don't have the formula.
Compare this to, say Bill James. In his revised Historical Abstract, he claimed that Biggio was the best player in baseball. This was a surprising conclusion, but you could tell exactly where all that value was coming from (e.g. HBPs, GIDPs, etc.) It conveyed real understanding to the reader.
Similarly, something like regression analysis of DVOA O/D/SP numbers versus scores (done by some other football site, I forget which) conveys real understanding of what the DVOA numbers mean.
Conversely, these predictions, because we can't reconcile our "model" of what the formula is with the numbers coming out of it, is not giving me any insight at all.
#150 by Eddo // Sep 11, 2009 - 4:13pm
That's a really good question, and I don't want to idly speculate why a defense would have a 14% drop (in the -16% to -2% scenario). I do know that the Outsiders have written that defensive success is more difficult to sustain year-to-year, according to their research, so I've always assumed that the defensive projections start with a much harsher regression to the mean than offense. Throw in that, if I recall, the league-wide defensive DVOA was -4.4% last year, and the regression for defenses like Pittsburgh and Baltimore might seem a bit much.
#152 by Bowl Game Anomaly // Sep 11, 2009 - 4:49pm
I only feel confident contributing a few small comments to this well reasoned and intelligent discussion: The first is that 8 years of data is far more than is relevant when predicting the future. I doubt there is any relevance to anything beyond the past 2 years when making predictions, 3 at the absolute most. The second is that whenever DVOA predicts a large drop, it usually relates to injuries, age, or both. Third, not only is defense more variable year-to-year than offense, but it is very rare for a team to sustain excellent defense for more than 2-3 years in a row (much more rare than you might expect).
All that said, I find it hard to believe that Pitt's D will fall as much as DVOA claims.
#155 by erniecohen // Sep 11, 2009 - 11:27pm
I'm glad you brought up the BAL D. Their last 8 years of DVOA read
-16.8, -7.6, -27.1, -20.5, -11.6, -23.6, -8.7, -24.5
and the 2009 prediction is ... -28.9
Yes, -28.9. The same system that expects the PIT D to collapse to their worst season of the decade (after their best year) simultaneously predicts the BAL D to exceed their best performance of the decade by a sizeable margin. Doesn't sound like regression to the mean, does it?
So what is the difference? Age? Both PIT and BAL D starters average 29. Is it that that the perf of the BAL D was being dragged down by Bart Scott and Rex Ryan?
Don't get me wrong - if I had to bet, I'd bet on the the BAL D to beat the PIT D in DVOA this year (albeit not by a large margin). But a 27% difference in their DVOA projections is just kind of bizarre.
#156 by Sifter // Sep 12, 2009 - 6:58am
The main difference between the two was injury rate last year. Baltimore (by the AGL) had the 2nd most defensive games lost, while Pittsburgh were 21st
The other thing that must be factoring for the Steelers is age - yes Baltimore has age issues too, but Pittsburgh seems worse to me. Look at their ages:
DL - Smith (33), Keisel (31), Hampton (32), Kirschke (35), Hoke (33), Eason (29)
LB - Harrison (31), Farrior (34)
DB - Polamalu (28), Clark (30), Taylor (29), Townsend (34)
The only guys under 28 are pretty unproven: Timmons, Woodley, Gay and then the rookies and special teamers. Only Woodley started last year. That's not a lot of youth there, so you'd think there is only one way the Steelers can go. And now they've got no Polamalu for a few weeks either.
Baltimore has a high average age too as you point out, but a lot of that comes from Trevor Pryce and Ray Lewis. By contrast they have some younger studs like Ngata and Suggs, the Steelers haven't uncovered theirs yet.
#158 by erniecohen // Sep 12, 2009 - 4:28pm
I agree about regression to the mean when it comes to injury luck. It would be nice to know how much of the expected regression is just because of this. (Of course the Polamalu injury is irrelevant in judging the prediction.)
The age model is somewhat more interesting, because except for Farrior (Townsend doesn't start), the PIT age issues are all on the DL, where they have good depth. In particular, the PIT DL don't depend on quickness, which makes them less suceptible to age decline.
I'm not so sure about the potential for improvement of the two teams, because I don't know the ages at which D players get less likely to improve. The PIT guys under 28 are 23,24,25; The BAL guys are 25,25,26,26,27,27. So BAL probably has a greater expectation of improvement, but I don't know how much. Again, it would be nice to know what the precise model is, e.g. so that we could make suitable refinements for the PIT DL situation.
#53 by bob (not verified) // Sep 09, 2009 - 10:48pm
Anybody have an explanation for why DVOA has a crush on the Jaguars? Every year we do the same song and dance about how "it's just what the numbers say" and then every year the Jaguars underperform deliciously.
#65 by Red Hedgehog // Sep 09, 2009 - 11:53pm
Except, of course, two years ago and four years ago when the Jaguars did quite well.
This information was given to me by my dog who, since the Eagles' signing of Michael Vick has become a Jaguars fan.
#54 by andrew // Sep 09, 2009 - 11:00pm
That's a very light yellow. On my screen, seperated from the non-playoff teams by the wild card blue, it isn't noticeable.
#81 by bubqr // Sep 10, 2009 - 6:16am
I've been wondering why the yellow looked like blue on my screen, and why teams with lower expected wins were division winners for a good 5 minutes.
#55 by Rover (not verified) // Sep 09, 2009 - 11:12pm
In case anyone cares, here are the largest disparities between DVOA and Vegas. Disparity comes by way of the W/L line, but also by the bet. For the hell of it, I added TMQ's prediction. (I hope I'm not violating some rules, but before websites like this, Vegas had the smartest analysts in all of sports.
Arizona: DVOA 5.5, Vegas 8.5 (under -125), TMQ 12
Buccaneers 7.1, (o +180), 4
Chiefs 7.6, 9 (o +110)
Houston 6.5, 8.5 (u -135)
Lions: 6.6, 4.5 (over -135), 2
Jaguars 10, 8 (over -135), 5
Rams 8.7, 5.5 (o -155), 5
Browns: 6.7, 7o (+175), 6 This one is actually very close, but the +175 creates the value.
Arizona, Lions and Jaguars are tough to take DVOA's side.
#100 by shake n bake // Sep 10, 2009 - 11:10am
Though when comparing anything vs Vegas you have to remember their ultimate goal isn't to be right. It's to make money off the bets placed on that line. That means either splitting the bets (so the losers pay the winners and the book takes home the juice from both) or (riskier) getting more of the money on wrong side. Both of these require the lines to be based off public perception as much if not more than what is actually the most accurate projection.
#160 by Jeff W (not verified) // Sep 13, 2009 - 1:06am
"Both of these require the lines to be based off public perception as much if not more than what is actually the most accurate projection."
In practice, they're pretty close to the most accurate line that "Vegas" can make by the time betting is over. The sharp bettors hammer errant lines back to normal.
#56 by Nathan Forster // Sep 09, 2009 - 11:14pm
As a fan of the Lions, I have to love that DVOA projects that 7-9 is more likely than 6-10. I'm just looking for that first win. Boy howdy, only 0.4 wins away from the Packers! I also have to say "kudos" to Aaron and the FO's for all of their great work here because I think the DVOA projections are top shelf.
I'm curious to what effect first-team (or any team) preseason performance has on preseason DVOA projections (if at all). I certainly understand the trends that are going against the Packers from reading my trusty copy of FOA 09, but it's hard to disregard what intuitively seems like powerful evidence that that offense will really cook (namely, Aaron Rodgers going dark Phoenix on all opponents). I'm wondering if this is something that's been looked at (although I wouldn't be shocked if it turned out to be a non-factor).
#97 by Bowl Game Anomaly // Sep 10, 2009 - 10:43am
I'm pretty sure that apart from injuries, preseason performance is not at all factored into the DVOA predictions.
#57 by JoeD (not verified) // Sep 09, 2009 - 11:17pm
Colts with 7th ranked special teams is pretty surprising. I'll take the under...
#79 by Ben // Sep 10, 2009 - 5:56am
I was thinking the same thing. I'd just be happy with 'Not gawd-awful' this year.
#60 by ChiTown (not verified) // Sep 09, 2009 - 11:35pm
I would say Pittsburgh looks too high, they look like an 8-8 team to me, and St. Louis is too low. The Steelers strugged with Dallas, got blown out by Tennessee, couldn't beat Indy, NYG, or Philly. They only went on a run (like the Cardinals) in the postseason. When KUBIAK has picked a team to come out of nowhere (TB, GB, CAR) over the years, the team seems to have everything go their way, so I'd put the Rams at 12-4. Also, Tennessee always seems to not be well liked by KUBIAK, so I'd put Tennessee at 11-5.
#103 by bengt (not verified) // Sep 10, 2009 - 11:47am
Let me help you rephrasing that:
The Steelers 'found a way to win' against Dallas, had a fourth quarter lead against Tennessee and were far from 'getting blown out', could have beaten Indy and NYG, and played like crap once against Philly.
#62 by Red Hedgehog // Sep 09, 2009 - 11:45pm
That 1.7% chance of a Super Bowl XXXVI rematch seems way too high for me. I guess that's because St. Louis' projection seems the least likely to me.
#63 by RaiderRefugee // Sep 09, 2009 - 11:47pm
How in God's name is the Raiders defense average (-1.3%). They could not stop the run at anytime and Drew Brees and the Saints embarassed the defense in week 3. Someone was chronic when these numbers were crunched. Otherwise, interesting and helpful.
#89 by Craig (N.S.) (not verified) // Sep 10, 2009 - 9:07am
#69 by shake n bake // Sep 10, 2009 - 1:23am
hmmm I'm puzzled by the projection for the Colts special teams. I'm optimistic for the new ST coach and the new punter McAfee looks like a real keeper (as a punter and kickoff man), but 7th? Maybe I've been conditioned to expect terrible special teams play from Indy, I'm not sure I can even picture what them having a ST unit in the top quarter of the league would look like.
#124 by Bobman // Sep 10, 2009 - 3:41pm
Yah, and I am a little concerned about a #1 O ranking despite some serious questions at LT. But it's possible. Maybe GiCharlie RobJohnson has been sending the FO staff Krispy Kremes round the clock....
More likely for the D and ST, though, would be a reversal of their projected rankings.
I wonder if that would be enough to move Indy to #1 overall....
#70 by AxDxMx (not verified) // Sep 10, 2009 - 1:29am
Wow.... The Chiefs are middle of the pack? That is beyond my wildest expectations. 2-14, here we come again!!!!!!
#71 by pachoo5 // Sep 10, 2009 - 1:45am
Wow, the one that really jumps out at me is the Falcons at 27th overall..give me a break. The Falcons will be worse than the Raiders, Buccaneers and the Lions? LOL, I think not.
It sounds like the Dolphins and Falcons are being way too heavily penalized for jumping from crappy to contending last year. Sure they might regress to the mean a bit but I doubt either team falls nearly as far as the DVOA projection is saying.
#72 by tally // Sep 10, 2009 - 2:05am
I suppose the Outsiders should take this as a compliment that DVOA is held to such a high standard of getting everything right. It's not enough to consistently beat the other prognosticators.
#109 by Ben Stuplisberger // Sep 10, 2009 - 12:50pm
I'd like to see evidence of that.
I take these predictions like I take all the predictions from every source, interesting, but ultimately worthless.
#154 by tally // Sep 11, 2009 - 6:40pm
I think it's King Kaufman at Salon.com who compiles projections from various pundits and at the end of the year and compares them to see which ones were the most accurate. FO has consistently been the top prognosticator.
#157 by Happy Fun Paul // Sep 12, 2009 - 11:22am
Alas, King Kaufman no longer writes a sports column at Salon.com. Perhaps someone else will take up the torch of testing the preseason predictions?
(I actually miss Kaufman's column quite a lot. Not only was it a fun read, but usually quite insightful. I can't recall for sure, but I think he was the one who tipped me onto the existence of FO, many years ago.)
#73 by pachoo5 // Sep 10, 2009 - 2:12am
The thing I question is, what is driving the huge jumps in wins for the Lions, Raiders, and Rams and the huge nosedive by the Dolphins and Falcons? I don't see any of those 5 teams going along with DVOA this year. Dolphins with the worst defense in the NFL and the Falcons with the 21st ranked offense???? What is causing the Rams defense to go from the 30th ranked defense last year to 9th this year? What exactly did they do on the defensive side of the ball this offseason?
#74 by pachoo5 // Sep 10, 2009 - 2:15am
The only big addition to defense the rams made this winter was the drafting of James Laurinaitis...this causes their defensive DVOA to jump from 30th to 9th???
#75 by afootballfan // Sep 10, 2009 - 2:15am
The beat the Ravens 3 times and the Chargers twice.
does that count for anything? lmao
#77 by Brendan Scolari // Sep 10, 2009 - 4:59am
I'm wondering, what are the reasons for the Niners defense falling from middle of the pack last year to 31st in the league this year?
They're basically the same team on defense except that corner Walt Harris is out for the year and Dashon Goldson is replacing the terrible Mark Roman at free safety. I know they were lucky on injuries last year, but I don't see them dropping too much.
#78 by Mat (not verified) // Sep 10, 2009 - 5:38am
I'm sorry but we all know the Dolphins and Falcons are going to fail this year. Last year was a fluke of easy schedule, ridiculous injury luck and sheer luck in other situations. A regression to the mean is ATLEAST respectable here considering the quarterback curse associated with a sophomore quarterback (Matt Ryan), and the Dolphins defense getting the injury bug and increase of age. And to add the Falcons having a harder schedule than last year all indicate the teams having a rough year. Not too often, in this case, have two teams been terrible in the prior year, rose to 10+ wins and maintained that the following season over the course of NFL history. I say the Falcons go 5-11 and the Dolphins 6-10.
#86 by Mr Shush // Sep 10, 2009 - 7:59am
Actually, I believe Lewin's research suggested that quarterbacks as a rule showed strong improvement from year one to year two (the other big jump being year four to year five).
I know Ryan's first year was so good as to make him something of a special case, and the Falcons did indeed have an easy schedule. I expect them to improve their already strong passing game, fall off significantly in rushing offense, and struggle horribly defending the run. If Abraham goes down (not unlikely) they will also have problems in pass defense. My guess would be 8-8, but nowhere between 6-10 and 10-6 would truly surprise me. 5-11 would.
I, incidentally, am a Texans fan, so there's no homerism at play here (though in fact I think the Falcons and Texans could be remarkably alike this season).
#82 by bubqr // Sep 10, 2009 - 6:21am
On that whole "DVOA effectiveness in predictions", is there any website/article out there (apart from Temo) running some numbers on who was the best in terms of predictions the past years ?
#90 by austin753 // Sep 10, 2009 - 9:11am
The Titans to have the worst special teams in the league? Has Rob Bironas' leg fallen off? I can't find any explanation in the almanac.
#95 by Tom Gower // Sep 10, 2009 - 10:20am
Have you seen Craig Hentrich kick lately? Have you seen them coverage? Have you seen the returners the past 6-8 years other than Carr and Pacman, neither of whom is on the team this season? Decent backup LBs are normally the mainstay of your special teams, and their LB backups are Keglar, Allred, and McRath, all of whom are young. I don't really think they'll end up #32, but if I had to guess I'd say they end up in the bottom quarter of the league, and, well, somebody has to be #32 in the preseason projections.
I'll also point out that Bironas, despite a strong leg, has never really had a down year of the sort all PK's have, and has missed FGs he normally makes in both playoff games, probably his two biggest/most valuable FG misses as a Titan.
#136 by austin753 // Sep 11, 2009 - 4:48am
Good points, well made.
After last night's shambles, -3.6% might be a tad optimistic.
#139 by JoeD (not verified) // Sep 11, 2009 - 9:36am
#93 by Jetspete // Sep 10, 2009 - 9:57am
I'm gonna assume DVOA is projecting offensive line injuries for the Jets this year, because with their running game and offensive line I cannot see them having the worst offense in the NFL.
#96 by Karl Cuba // Sep 10, 2009 - 10:33am
OK, slate the niners all you want but I really can't see the Rams and the Seahawks as the 11th and 12th best teams in the NFL.
Thae Rams are pretty dire at nearly every spot, I just don't see it.
I don't see why so many pundits, including several from this site, keep saying, "why would this team have forgotten how to win?" But this isn't the same team, who's left from their superbowl run? Aging versions of Jones and Hassleback on offense and a pair of linebackers. Jones is already banged up and with him out of action there's a good chance Hassleback will get dinged too.
I don't really see the niners D slipping to 31st either has Aaron done something to Willis?
#99 by Scrapper (not verified) // Sep 10, 2009 - 11:03am
Aaron states in the prefatory commments that "There's no reason to really regurgitate the reasons why our projections for some teams (Jacksonville, Green Bay, the entire NFC West) differ so greatly from the conventional wisdom -- we've gone over it in numerous articles and in numerous promotional interviews over the past few weeks.".... While that in fact may be the case, Aaron, many readers may not have not found these (unidentified) articles and would like to hear more about the rationale for these provocative picks. If you want to point us to the specific articles and promotional interviews, we can all go to these other sites but it might just be easier to address these points at your site. Thanks for the great work!!
#104 by anonyomaomgdjdsgk (not verified) // Sep 10, 2009 - 11:54am
Seriously, I don't see these explanatory articles linked anywhere on here. The "players to watch" Insider articles certainly don't provide any illumination.
#106 by Jimmy // Sep 10, 2009 - 12:22pm
If you look for 'FOA Media' in the Extra Points archive you will probably find some information about whichever of the projections annoys you. Or you could purchase their book.
#105 by Unanimous (not verified) // Sep 10, 2009 - 12:15pm
This site is dead; it served its purpose. If you really want to know what Aaron thinks, it'll cost you.
#117 by Rover (not verified) // Sep 10, 2009 - 3:07pm
What, you want your money back? Aaron's not allowed to make a living?
#111 by JasonC23 // Sep 10, 2009 - 2:01pm
The negativity and cynicism in these comments are disheartening.
#113 by D // Sep 10, 2009 - 2:26pm
I'll just say that the first thing that pops out to me is Indy have the 7th best special teams in the league. The Colts having awful special teams has been one of the most consistent things in the NFL over the years.
#128 by shake n bake // Sep 10, 2009 - 4:21pm
Improvement is probably likely with a new ST coach, the new punter looking like a stud, backup LBs with pro ST experience and Indy actually signing a veteran STs coverage man (Francisco from 'Zona), but 7th seems like a bit much. I'd be absolutely thrilled if they had a ST unit that not only wasn't cringe inducing, but a net positive. I won't be expecting it from them though.
#118 by Unanimous (not verified) // Sep 10, 2009 - 3:10pm
The negativity started with Aaron's comment that he's "only going to say this once." I'm sorry that he has so much contempt for his readers. Rather than express surprise at what are some very surprising things in this report, are we expected to puff on our pipes, tug on our beards and mutter, "hmm. . . very interesting"? Or maybe he expects us to lap it up and grovel in supplication at his magnificent gift to the world. Every year there are surprises in DVOA, and every year people here comment about it, and every year Aaron or someone else tells people how stupid they are for being so "subjective" and "irrational." And every year Aaron and co. are wrong about a helluva lot of things. Put that on the masthead.
#133 by anotherpatsfan // Sep 10, 2009 - 5:54pm
Bullshit. Why don't you actually read what he said:
"I'm only going to break in once to remind people: This is not "Aaron thinks." This is "the numbers think." I don't think the Steelers defense will be average, and I don't think the Jaguars defense will be this good. These are statistical projections based on a formula -- the best formula, we believe, but not a perfect one, and not one that takes into account my subjective opinions.
These numbers also represent the average of a range of possibilities. All the subjective issues that we know play a role in wins and losses -- teams being healthier or unhealthier than average, young units suddenly gelling with experience, Mark Sanchez being the second coming of Matt Ryan, players attacking their MTV hostess girlfriends, etc. -- create that range of highs and lows."
Contrary to your rant, Aaron said he is only going to interrupt the legitimate poster/troll discourse once to remind people he is not giving his own opinions, but just setting out numbers generated by an admittedly imperfect formula -- a formula which does apparently take into account statistical/game chart data which he deems important, but which doesn't take into account his own subjective opinions that disagree with the output from FO's formula (i.e., he thinks DVOA will get things wrong, as with the Steelers defense example).
He didn't say anyone was stupid for having subjective opinions at odds with the DVOA numbers (and I think every functional football fan would have such opinions), he just presented the DVOA numbers generated by FO. He didn't say don't comment about DVOA in the negative -- he just put this info up for discussion. Disagree with the DVOA results all you want -- there are many potential subjects of argument.
Yes, despite the fact FO gives away a bunch of free content (like this freaking article), they charge for the right to look at some of the numbers they generate. Yes, the DVOA-generator is a bit of a black box KFC-seven-herbs-and-spices secret formula. Get over it. FO is a business, and is entitled to charge for the numbers or anything else they think someone will buy; we as consumers of this content (again, must of which is free) are entitled to buy the extras, or not. FO is also entitled to keep the nuts and bolts of the DVOA formula a secret, as that formula represents Aaron's/FO's intellectual property, which most businesses don't just give away. If all that is too much for some posters here to deal with, just click on back to the Fox message boards. Please.
#119 by Jesus H. Favre (not verified) // Sep 10, 2009 - 3:18pm
So if past Eagles teams, good and bad, were actually awesome according to DVOA how awful must this year's squad be? If we do the standard "chop 25%" off their rating, they're a bottom-10 team.