Where our work appears on mainstream media outlets
Top 35 quarterbacks charted for 2015 with scouting notes. PDF only $9.99
Click here to buy PDF version
Like our page on Facebook and get Football Outsiders links directly in your Facebook feed.
Official Account: @fboutsiders
Scott Kacsmar: @FO_ScottKacsmar
Ben Muth: @FO_WordofMuth
Aaron Schatz: @FO_ASchatz
Vince Verhei: @FO_VVerhei
-- plus --
Bill Connelly: @SBN_BillC
Cian Fahey: @Cianaf
Brian Fremeau: @bcfremeau
Tom Gower: @ThomasGower
Andrew Healy: @AndHealy
Rivers McCown: @RiversMcCown
Chad Peltier: @CGPeltier
Rob Weintraub: @robwein
Sterling Xie: @SterlingXie
Need help on FanDuel or DraftKings? Purchase a subscription from the website that specializes in daily fantasy.
21 Sep 2006
Here's our look at all the big games in a week filled with big games. This week featuring French and more commentary destined to turn Vikings fans into this year's Falcons fans.
Posted by: Mike Tanier on 21 Sep 2006
130 comments, Last at
25 Sep 2006, 6:21pm by
Is the point of the article to highlight the hype of the media (Games you will watch) vs. the actual good games on the schedule (Games you should watch)?
1. Nope. More like "The big games that have to go first" followed by "Games I feel like writing more about." Came up with the categories last August. Haven't come up with better ones yet. I led with Saints because a) U2 concert/visit from ex president and b) I haven't written more than a nutshell about them in about a year.
Wait, where did you hear that Dexter Jackson and Rich Braham are out for the year?
Last word out of Cincinnati was that both will be out for a few weeks, but will return. Losing them both for the season would be disastrous.
Also of note regarding the Bengals injury woes, kick returner Tab Perry is out with a hip injury.
I wonder how close the 10 points in MIA-TEN is to the most ever in a game between two winless teams, week 3 or later.
"This game is an Ed Wood movie; Favre is Bela Legosi."
That's probably the most apt definition of this year's Packers that I've yet seen.
Hey, I wouldn't risk any geld on the Scandavanian raiders going against the inexplicably productive ursine ones on Sunday, so don't accuse me of being like last year's Home Depot homers.
Then again, if the money line gets tasty enough..........
"If Favre weren't playing, who would you tune in to watch? Ahman Green? Shaun Rogers? A.J. Hawk? Mark Furrey?"
The Lions are so nondescript and unknown that when you point out this fact by mentioning one of their unknown players, his first name is incorrectly identified. It's Mike Furrey.
"Eagles fans: Still have that cold feeling in the pit in your stomach?"
Yes. Yes, I do.
"But as you probably know, the official NFL rankings aren't worth the bandwidth it takes to download them. The NFL ranks teams by yards gained and yards allowed. If the Bears roll up a big lead and then slip into their prevent defense and allow easy yards, the official rankings don't care."
Uhm... the NFL ranks them 5th, while DVOA ranks them 7th... so I don't think you can really use that example.
Mike, Props for knowing that Monday is a Red Beans and Rice day (I won't get into the history).
I have to admit that I'm more pumped about Monday night than I've been for any other NFL game ever. This week has dragged on for me, but it's so, so great to have my teams back in the Dome.
Should be "les bons temps"; adjectives accord in French.
Sorry, I don't want to turn this into another grammar bores thread, but if you're going to go all multilingual on us...
I would like to nominate No pirate -themed team has scored a touchdown yet this season as statistic of the year.
I would also like to express my appreciation for concluding Leinart funny.
Is there a FO thread to vote on Mike's best one-liners? May be time to start one.
My nominations are:
"If you thought Tom Brady and Jake Plummer would meet in this year's AFC championship game, you were a little off target. But you weren't as off target as Brady and Plummer have been in the last few weeks."
"Apparently, the Titans have a drill in which they play football for 30 minutes and only pick up one first down."
the above-refeenced Bela Lugosi line.
Way to go, Mike. Though I would think the Indy/Jax game would rate higher--it could be either the second-act of the coming out party of a new powerhouse, or prove that Indy is NOT the Indy of 1999-2004 all over again. Either way, should be relatively close and have post-season implications.
How can something be opaque and translucent? Oh wait, is this an exercise to give myself a headache?
Anyone, from Baltimore know the skinny on the Ravens injuires. Fox has Aubrayo Franklin, Ray Lewis, JAmal Anderson, Haloti Ngata, Jonathan Ogden,Aldaius Thomas,and Samari Rolle, all questionable for Sunday. That's like half the team, or half the good players anyway. ch
Can you really have a coming-out party with a team as old as JAX?
"Somebody told me the reason we came back was because the Eagles were tired from kicking our butt in the first half," Michael Strahan said after the game.
Is that why they switched to kicking the Giants' crotch?
16: Some people have coming out parties well into their 30s. Although those parties are usually considerably gayer.
"Can you really have a coming-out party with a team as old as JAX?"
You could if Esera Tuaolo played for the Jaguars.
Some people have coming out parties well into their 30s. Although those parties are usually considerably gayer.
Hmm... muscular, sweaty men dressed in bright colors chasing balls, piling on top of eachother, and slapping eachother's asses. Yeah, WAY gayer.
Billick loves to label players questionable even though its almost assured that they are going to play. I'm pretty sure everyone you mentioned except Franklin is going to play, and Jamal Anderson is definitely not going to suit up.
Billick isn't the only one.
And in other news, I heard that Tom Brady was only probable this week, with a shoulder injury...
Apparently, if I don't accept changes in Word after Tim edits Rundown, then both my mistakes and his corrections make it into Rundown. That explains the opaque thing. That also explains last week's grammar problems. At least that mystery is solved.
The pirate-themed teams not scoring touchdowns stat is debatable. I'd say that the Vikings are essentially pirate-themed, and they've scored touchdowns.
From what I've heard, Dexter Jackson is just out for a few weeks, and there's a strong rumor that Rich Braham is out for an extended period of time, possibly the year. But due to Braham's age, we already had a groomed backup waiting in the wings.
Pollack either didn't play or barely played in the preseason and the first two regular season games; since he hasn't been much of a factor, thus far, I doubt we'll see a dropoff, because of his absence. (Obviously, we'd be much better with him playing, though.) Likewise, for as much hype as Odell Thurman has gotten, he's had serious problems with overpursuing, and starter Brian Simmons moved over a slot to fill in for him--he's arguably been doing a better job than Odell. Our defensive coordinator actually implied that he'd rather have Simmons in there, anyway, as Odell is a bit...troublesome.
For even more injury fun, there's our WRs. Of our eight WRs (we carried seven, and had to move one up from the practice squad--Reggie McNeal, former scrambler QB, we may use him on kickoffs), all but three (Chad, Washington, and McNeal) are dealing with varying degrees of injury. And Chad obviously got jacked up at the end of the Cleveland game, so that may factor in, as well. I'm hoping that TJ and Chris Henry (who, despite his off-the-field problems, has been a great player) will be able to play, and it looks like they will. I'd be happy with two of our top three WRs, though.
I think we'll be better down the stretch, this year, as the moderately-injured players come back...but as for Sunday's game, I've no idea. I can see it going either way.
I'm not so sure that the Bengals aren't missing Odell.
Gonzalez - 10 rec for 81 yards
L Johnson - 5 for 80
Winslow - 4 for 42
Droughns - 3 for 30
Harrison - 2 for 17
Heiden - 2 for 13
Their opponents have been avoiding the WRs and targeting the TEs and RBs so far this year.
Vikings are not pirates.
This game is an Ed Wood movie; Favre is Bela Legosi.
#26--even with Odell in there, we've had a problem with TEs. You may very well be right, but overall, I've been happier with Simmons' performance. Though I'm sure that once Odell gets more experience, he'll be far better than Simmons currently is.
I see no one will address the issue of Defenses as long as it's clear how awful your random stats are.
If you want any credibility besides how good you guys are at wanking yourselves, why don't you explain the why the Bears are #7 and the Ravens are 2x as good as the Bolts, even though they've done about the same thing.
re 29, 26, etc.etc Wow, good thing they still play these things between the white lines... wait a minute, they DO still play on SUNDAY between the white liney things right?? BTW, last time I checked, the Bengals O was capable of running up the score and the D that will show up on Sunday is still better than the D from LAST year. Steelers sure struggled to put up points on Monday night... Sure they might put up more vs the Bengals D, but 20 points more? Bengals will still score around 30.
who picked the swagger caption for the ray lewis picture? was that intentionally funny?
Per Aaron Wilson of Ravens24x7.com, all of those players are expected to suit up and play on Sunday except for A. Franklin and possibly Musa Smith.
You got that right. Lots of coaches are gaming the injury list now. Or maybe they've always done it, and I'm just noticing it more now. Anyway, lots of coaches do this. Fisher is notorious for it. Belichick's injury reports should be eligible for short fiction awards. The Colts had 20 guys listed as Questionable last week. Want to guess how many actually missed the game? I think it's likely the league will try to come up with a something in the offseason to tighten up or change the rules and make the whole thing more meaningful.
I don't think you will get any complaints from Will or I about calling the Vikings a .500 team. That was our position from the get go I thought :)
We took issues with the 5 or 6 win team thesis.
I do think they have a decent chance vs the bears though, not good, but decent. If they don't win this game though they aren't making the playoffs.
Vikings were Scandanavian pirates and raiders.
#33: Explain why Chicago is 7th and Baltimore and SD are 1 and 2 (with the Ravens being 2x as good as the Bolts).
Does that sound like a "reasonable measure" to you? Not to me.
Because San Diego and Baltimore have played Oakland, and the stats aren't adjusted for opponent yet.
That's just a prediction, though. 8 wins is certainly not far from 5-6. So far the coach has looked solid, the QB has been healthy, and so all of the variables that likely pull things downward because there's just so much potential for things to go wrong haven't blown up yet.
Plus it's not like those wins were terribly solid, either. If those two wins end up getting them to 8-8, that's not too different than a 5-6 win team.
8 isnt far from 5 or 6? what are you talking about pat? A 2/3 margin of error in a realistically 8 or so range is atrocious. Using that logic no team is far from anything. Hell the Colts last year weren't too far from missing the playoffs. :)
Now to a certain extent this is probably true: Luck, Injuries, Scheduling, Whether etc. all make any projection a crap shoot, but lets not pretend a typical 8 win team is likea typical 5 win team.
I typically envision the league breaking into 5 groups each year:
0-4 Horrible teams (Free wins)
5-6 Bad teams (Occasionally Surprise people)
7-9 League Average (Inconsistent for whatever reason, or just not talented enough)
10-11 Solid Teams
12+ Dominant teams
Being a 6 win team is different from being a 8 win team often. Granted there are the vagaries of scheduling and such, but an 8 win team usually has to beat someone decent once or twice. Anyway, bottom line is I think the Vikings are a lot more likely to win 9 then they are to win 5 (and I was saying that before week 1). Given their start I think they are looking at 8.5 or so now, and 10 isn't out of the question if they can manage to win this week (I am not super confident though, CHI looks amazing).
It is true that they were both ugly wins, but they did come against what will probably be average teams.
Anyway, I love FO and cannot wait until we get the more robust ratings. I look forward tothe season so much and my loyalty is 95% to the NFL and 5% to MIN. I am guessing they hover in the middle third of DVOA all years.
No, sorry. Only a poor statistic can provide such a ludicrous spread... but your boys won't confront that... they'll pretend all is well. Lalalalalala.
The 49ers will pull the upset this week against the eagles!!!!
Well fresh, if you really want to see how they come up with the statistics, you could click on the button at the top, "just the stats". However, since your first post included the phrase "wanking yourselves", I'm betting that you are happier just being ignorant.
"Brady's Patriots are 2-0, thanks to effective defense, a solid 1-2 punch at running back, and awful opponents."
So it's common knowledge now that the Jets and Bills are "awful?" This is a superficial throwaway line and not worthy of FO in general.
I think it's a solid assumption that the Jets are going to be a bad team. They have a bottom 5 run game, a spare parts defense that lost John Abraham and Ty Law and is trying to play the 3-4 without a nose tackle, and they aren't going to be able to come back from deficits because they play a short pass game, despite the huge YAC efforts of their receivers this past Sunday. Awful? With some bad luck they could be awful. I think they're more likely headed for 5-6 Wins, provided Pennington stays healthy. I've always liked him, but he's not a QB who can counterweight the mediocrity all over the rest of the roster.
The Bills have more going for them, in my opinion. Their D has the potential to be good, and that means they'll have a chance every week. The run game is okay, but I don't think the words "Losman fires it to Peerless Price!" inspire fear in anyone. I see them fighting for .500 unless their pass game just utterly handicaps them.
If we moved Pennington and Coles over to the Bills roster, I think we might have something. That could be a team that competes for a playoff spot.
There were some Bengals fans on this thread earlier, wondering how they feel about the Bengals calling pass plays with a few minutes left in a game they led by 24 points? With the starters still in the game? With Chad Johnson getting injured on the play?
OK, since I'm guessing you want this to be justified by only looking at "traditional" stats, I created a little formula to make 1 single "defense rating" just by adding together the good things a defense does, and subtracting or dividing by the bad things. I did NOT make this formula, then tweak it until I got the results, I just thought up the formula, calculated it, and am listing the results.
So here is the formula:
DEFENSIVE RATING = COUNTING / RATE
COUNTING = 100 * ( 4*[TO] + 2*[sacks] + [inc pass] - [1st downs] ) / [total plays]
RATE = [avg yds per pass] + [avg yards per run]
So basically, it's good plays minus bad plays, adjusted a little for the importance of turnovers, and for how many plays they've run. Then that is divided by [yds/pass + yds/rush]
Here are the results (hope this doesn't look too ugly):
Baltimore Ravens â€¦ 9.82
San Diego Chargers â€¦ 5.5
Seattle Seahawks â€¦ 4.57
Atlanta Falcons â€¦ 4.37
Jacksonville Jaguars â€¦ 3.94
New Orleans Saints â€¦ 3.15
Chicago Bears â€¦ 3.06
Cincinnati Bengals â€¦ 2.48
Philadelphia Eagles â€¦ 2.24
Buffalo Bills â€¦ 2.07
Pittsburgh Steelers â€¦ 1.85
St. Louis Rams â€¦ 1.82
San Francisco 49ers â€¦ 1.71
Dallas Cowboys â€¦ 1.59
New York Jets â€¦ 1.41
Green Bay Packers â€¦ 1.4
New England Patriots â€¦ 1.36
Washington Redskins â€¦ 1.35
Arizona Cardinals â€¦ 1.32
Tampa Bay Buccaneers â€¦ 0.99
Minnesota Vikings â€¦ 0.85
Denver Broncos â€¦ 0.58
Cleveland Browns â€¦ 0.5
Carolina Panthers â€¦ 0.32
Indianapolis Colts â€¦ 0.14
New York Giants â€¦ 0.07
Kansas City Chiefs â€¦ 0
Oakland Raiders â€¦ 0
Miami Dolphins â€¦ -0.09
Detroit Lions â€¦ -0.15
Tennessee Titans â€¦ -0.69
Houston Texans â€¦ -0.97
If you'll notice, Baltimore is 1st, almost twice as good as SD, who is in 2nd. Chicago is in 7th. I had NO IDEA this was gonna mimic VOA so well. Maybe later if I have time I'll take a closer look at these numbers. But for now I'd say the onus is on you to come up with some reason why the 3 teams you mentioned SHOULDN'T be ranked the way VOA has them.
8 isnt far from 5 or 6? what are you talking about pat? A 2/3 margin of error in a realistically 8 or so range is atrocious.
The DVOA prediction for Minnesota was 5.9+/-3.54 wins. 8 wins is absolutely in that error margin.
I think you're envisioning some strange predictive world where this isn't a game that they play. You're never going to get better than 50% accuracy.
Oh, let's see if I can come up with a stat that shows something different... hmmm... oh here we go...
...I didn't even need a slide rule!
If you actually had any clue whatsoever as to how the Estimated Wins totals came about, you'd be embarrassed. So I'll just feel embarrassed for you.
Assuming Fresh can follow an equation that "complicated" is going out on a bit of a limb.
I know I'm going to regret this, but what in the hell are those numbers supposed to mean?
Well all I am really saying is that I thought 5.9 was low and I think MIN is an 8 or so win team (possible more since they already won 2 of their toughest games). To which your reply is that 5.9 wins is more or less 8.
My gut tells me most seasons I could pick say 30 teams records within a +/- 3.5 wins. Perhaps I am wrong about that I would need to do some more statical studying.
In fact looking back on my preseason predictions I would be amazed if I didn't get 30 of them within that margin. Right now the only two real threats look like TB and ATL.
Like i said I could be all wrong about this, but my inuition when someone says I expect X to be a 6 win team and they end up an 8 or 9 win team is not that they were correct. Not knocking FO, as I have said many times I think its great.
Like Will though I just think the preseason predictions were missing something with MIN (IMO easy schedule combined with large homefield advantage and improved coaching).
49: You can run all the numbers and pump all the spreadsheets you like... if a stat is divorced from reality its worthless.
You do know what stats are, right? It's an attempt to quantitate the qualitative. All they can EVER say is "These two things are different" DVOA or whatever is only as useful as it matches what you see with your own eyes.
Hey, it may be catching something that I don't see... I'm prefectly willing to admit that, as long as you can explain why Baltimore is 2x as good as SD and Chicago is 7th without saying "the numbers work out that way".
50: That would be points allowed this season... a pretty telling stat in regards to wins and losses, but I didn't need a calculator to add up 2 field goals or 1 touchdown.
They aren't all predicted with a +/- 3.5 win margin. Minnesota's standard deviation is one of the largest, trailing only New Orleans and St. Louis.
Like i said I could be all wrong about this, but my inuition when someone says I expect X to be a 6 win team and they end up an 8 or 9 win team is not that they were correct.
No offense, but if that's true, your intuition is flat out dead wrong. It says it right there: 5.9 +/- 3.54. 8 is entirely consistent with that. 12-14 wins , or 1-2 wins, would not be.
The simulation is flat out telling you "hey, the average number of wins was 5.9, but I'm really not sure about it." And now you're complaining that it might be off by 2 wins? It told you that it could easily be off by that much!
as long as you can explain why Baltimore is 2x as good as SD and Chicago is 7th without saying
Because Baltimore and SD have played Oakland, and Chicago has not. And Tampa Bay actually looks like a significantly worse offense than Tennessee.
If it was possible for teams to score negative points, Oakland would've. Green Bay at least put up a fight. Oakland was beyond awful.
I believe those are the Points Allowed totals for those teams.
55: That's not much of an analysis - "Blame Oakland"
I'm a Ravens' fan, so I'm not hear to cheer about Chicago... but how can a team that dominates 2 straight games and only allows a single TD in that span be in 7th in regards to defense?
What are you saying with that? That they were *lucky* not to give up more points?
That doesn't match what I saw.
That would be points allowed this seasonâ€¦
Points allowed isn't a good measure of a defense, just like points scored isn't a good measure of an offense. An interception returned for a touchdown isn't the fault of a defense, and a defense shouldn't get credit if the offense is going so well the other team never gets on the field.
In total DVOA (offense plus defense plus special teams), SD and Baltimore are essentially tied, and Chicago is 3rd. Does that make you happy?
You're assigning too much to Chicago's defense and not giving their offense enough credit.
55: Thatâ€™s not much of an analysis - â€œBlame Oaklandâ€?
Have you seen Oakland play? If you believed that Baltimore and San Diego were responsible for those offensive performances, you would've thought they were utter and complete gods.
that dominates 2 straight games and only allows a single TD in that span be in 7th in regards to defense?
You're not giving nearly enough credit to their offense, which never let Green Bay start in good field position. Pretty much the same for Detroit (although it happened once). Good offenses make defenses look better.
58: Obviously there is an interaction between offense and defense... but I am looking solely at D here... you seriously want to tell me that Chicago is the 7th best D you've seen so far this year?
I just wanted to thank you. It's a rare occurance that someone will start off posting as a typical troll and then come back with a reasonable well thought out response. I'll have to go look at the stat lines and play-by-play for the games, but I'll try to give you an answer.
For now, I'll just say this. Until we start getting Defense-Adjusted numbers (I think that starts after Week 4) the column you want to look at is the "FOX RATING" numbers. Those numbers take into account the pre-season projection for each team in hope of combating out-lying data due to limited sample size (2 games isn't enough to make a conclusion about anything). And the defensive FOX RATINGS have Baltimore #1 (-20.9%), Chicago #2 (-15.7%), and SD #3 (-14.0%).
59: Yes I've seen Oakland play... twice, unfortunately.
So are you saying that Chicago's D is ranked "normal" and SD and Baltimore are in the stratosphere because they had the pleasure of playing the Raiders?
I can accept that to some degree... however it doesn't explain the teams that didn't play Oakland that are also in front of the Bears.
Nobody is actually saying that Baltimore is twice as good as SD. All that DVOA measures is how much better or worse than average, on a per play basis, a team has performed SO FAR. And, if you look at ALL THE PLAYS, not just the 4 scoring plays for these teams, Baltimore has been better. They've forced more turnovers, allowed the fewest yards per pass and yards per rush (by a HUGE margin), given up the fewest 1st downs. I haven't counted, but I would bet they've forced the most 3 and outs.
This is definitely NOT the same as saying Baltimore is the best. The early season numbers aren't adjusted for opponent, Baltimore will surely come down after you take into account who they've played. More importantly, 2 games is a small sample size, and there is a lot wider spread of values now than there will be after teams have played a few more games. Baltimore is not going to keep performing at this level, and they'll come down to the pack.
Oh, and on your chart, why don't you count the points the defense has scored, as well. Then it's
A couple of things:
Don't feed the trolls.
I'm puzzled by this Vikings/Bears thing.... FO is willing to use VOA to show why the Bears are much better than the Vikings, despite the 2-0 records (which, like Becephalus and Will, I don't necessarily disagree with).
BUT, for teams that you don't believe are as good as their VOA indicates, you are perfectly willing to write off their VOA and record as products of the opponents that they've played (for example: "Of course, the Saints' 2-0 start is as much a product of the schedule as any improvement on their part.")
So why don't you point out that we really don't know if the Bears (and particularly their offense) are any good because they've also played 2 horrible teams? Why are they different from the Saints? Obviously, at this point in the season there haven't been enough games to really know for sure who's good and who's not. So FO is using gut feelings and whatnot to fill in the void.
Which is okay, but you should probably be more open about it. It just bothers me to see you say (I'm putting words in your mouth here): "Chicago is good because they have a good VOA (who cares who they've played?). Minnesota is average because they have an average VOA. The Saints are bad because they've played bad teams, VOA be damned."
but I am looking solely at D hereâ€¦
No, you're not. You're trying to justify your subjective opinion of Chicago's defense by looking at the amount of points that Chicago has allowed, while ignoring the fact that Chicago's offense has been fantastic and put Chicago's defense in easier spots than, say, Seattle and Jacksonville.
You can't say "Chicago's D is obviously just as good. They've only allowed 7 points!" Points allowed != defense.
I can accept that to some degreeâ€¦ however it doesnâ€™t explain the teams that didnâ€™t play Oakland that are also in front of the Bears.
Seattle, Jacksonville, Atlanta, and Cincinnati's defense have all been put in far worse spots than Chicago's defense has.
Chicago's defense hasn't really had the chance to prove itself dominating yet. Hasn't needed to, because their offense has simply ripped the other teams apart.
It appear to me, that you all are penalizing the Bears' D because their offense has been remarkably competent...
Am I the only one who think that makes *zero* sense?
Why is it impossible that you're giving the Chicago defense too much credit because their offense is competent?
First, I wanna agree with Wanker. Thanks for actually listening and responding reasonably.
Second, I would guess most of us haven't seen all the teams yet this year. I certainly haven't. Sure, I've seen highlight packages, but they don't do a good job of telling the whole story (usually). The only thing I have to go on is the record of what happened during the games. And since I also haven't had the time to go over the game logs for every single game and really inspect them, I'm just kind of trusting VOA (based on my past experiences the last 2 years) to do a reasonable job of interpreting the action. But I'll go over the play by play from some of the teams ranked above Chicago and see whether they seem better or worse. Are there any that you specifically think should be ranked lower?
65: I've got a newsflash for you... statistics are subjective.
What makes you think X should be divided by Y to give you a "better" assesment of what's going on? Your subjective judgement.
Statistics are not subjective. You measure the quality of a statistic by how well it predicts the future.
What makes you think X should be divided by Y to give you a â€œbetterâ€? assesment of whatâ€™s going on?
If it predicts future behavior better, it's better. This is not subjective. It's objective.
What makes you think X should be divided by Y to give you a â€œbetterâ€? assesment of whatâ€™s going on?
Because every time the staff makes a change to the way DVOA is calculated, they look at how the correlation to future wins changes. If the correlations get better, they keep the change. Worse, they drop it. The weighting of different elements is determined the same way. Whatever weight results in the best correlation is the one they use.
So yeah, I guess it's subjective in the sense that they are choosing "wins" as the most important stat.
[I might be wrong about "future wins" being what they correlate it to. Can't remember off the top of my head. All I remember is the word "wins" in there somewhere.]
68: Honestly, I haven't seen the other D's play either... except for Jax on Monday night, and I am perfectly willing to admit that they are playing "lights out" and thus should be ahead of Chicago... but the Bengals and Seahawks?
And if there is any team whose D should be penalized for it's offense, then it's the Falcons. I've only seen a bit of them, but their gajillion yards rushing is going to make it easy for Larry, Moe, and Curley to look like Pro-bowlers.
If I was going to ask for anything (which I'm not really) I'd like to know why the Falcons D is better than the Bears.
It's not that I don't realize that the Falcons are playing good D... but I don't think they're playing better than the Bears... why am I wrong?
70: What's the R-squared over a season? I bet it sucks, or otherwise these Outsiders would be billionaires.
Keep in mind Tampa is also very close to Oakland-level stink fest. That's why Baltimore's so freaking high - because it played both. That's where a lot of Atlanta's benefit comes from.
Right now Chicago's also at a bit of a disadvantage because the teams that it played aren't as god-awful bad on offense as the others.
That'll change once the opponent adjustments come in.
I'm not a billionaire, but I had a pretty profitable season last year using DVOA as my main tool.
The correlation is actually to points, not to wins. Estimated wins correlates to wins.
I bet it sucks
What is it with armchair statisticians who don't understand what r-squared actually means?
R-squared is a measure of how well the statistic you're generating explains the spread you see. You are never going to explain all of it.. It's a game. A low correlation coefficient doesn't mean that the statistic is meaningless. It means there are other factors involved - like luck or injuries, for instance.
If you want to know if something "sucks" or not, you use p-value. C'mon. This is basic statistics here.
75: There are two things there... over the season it presumably performs better than betting on the home team or whatever. Great.
However, isn't the purported strength of this stat the ability to work with individual plays?
What are my confidence intervals with that exactly? I'm supposed to believe that because a magic number does marginally better than Vegas every derrived stat from it is gold?
oh, and the correlation is definitely better than the one for points scored.
I am searching for the articles explaining this, but I can't find them? Links anyone?
Aaorn et al - One useful thing that the site needs is a better explanation of WHY we all think DVOA is useful. There is an explanation of how it is calculated (kind of), but it'd be nice to have an easy place to link to for situations like this - a page with all the various correlation coefficients for the newest versions of all the stats.
76: I am not a statistician, though I use them everyday...
You just said that prediction is what proves how good a stat is, then you said that a low R isn't a big deal.
I think you may need to re-read your stats book.
You just said that prediction is what proves how good a stat is, then you said that a low R isnâ€™t a big deal.
Yeah? So? There's no problem with those two statements.
You tune a statistic by maximizing the correlation coefficient. But if the correlation coefficient isn't incredibly high, that just means that there are other factors at play.
78: I defintely agree. I'm obviously a newb, and I've looked at how it's calculated and it seems reasonable... however, when it diverges from what I think is right I want to know *WHY* and why I should trust DVOA instead of anything else.
I'll say right now that I thinkpoints scored against through week 2 is a bette predictor of how D's will finish than DVOA says.
It's right here, for the previous version of DVOA.
The correlation (r) to points is higher than 0.9. The correlation to wins is higher than 0.8.
80: Have you tested it versus noise?
Iâ€™ll say right now that I thinkpoints scored against through week 2 is a bette predictor of how Dâ€™s will finish than DVOA says.
Send Aaron a message via the contact form. Ask politely if points allowed through week 2 is a better predictor of points allowed through week 16 than defensive VOA is.
82: Yeah, that's pretty special. Show me a Bland-Altman, and we'll see for sure.
80: Have you tested it versus noise?
That is what a p-value does.
I didn't have anything to do with testing or coming up with DVOA. I do have a lot of experience with statistics, though. And what Aaron's doing is perfectly backed up.
OK, so I took a closer look at the drive logs from the teams ranked above Chicago, and I'm starting to see why Fresh has a problem with VOA. Take, for example, length of drives. Here are the teams, along with how many drives went 30+/40+/60+ yards, then the last # is 3&outs (or TO before 3 plays):
Has anyone ever looked at whether teams can be consistently good or bad at keeping the other team from stringing together first down. What I mean is, given 2 teams that allow the same number of first downs, one of them might allow a long scoring drive, then a bunch of 3 and outs, while the other might allow exactly 1 first down per drive, but no longer drives. Has anyone studied whether this pattern or ability is consistent, or whether it is just random. Next game (or season), will these teams show the same patterns, or are they just as likely to be reversed?
Bland-Altman plots would be highly biased in this case, since you're tuning the statistic.
Thanks for the link Pat, although it's not nearly as convincing as I remember. I swear I have seen charts that show the correlation to future wins, or maybe future points, (rest of season? next season? I don't remember) for DVOA, wins, points, etc, and show DVOA performing the best. But maybe that is something I constructed in my head to ffol myself into trusting the numbers more than I should.
86: Oh no. I'm pretty sure your model has blown every single assumption that goes into a p-value out of the water. You need to generate some white noise with the propper value range and run it through and see what you get.
88: I would guess a huge proportional bias... but you should be able to do a Bland-Altman, unless your stats are nothing more than an autoregressive curve fit.
87: I appreciate you doing that... very interesting.
You need to generate some white noise with the propper value range and run it through and see what you get.
Sticking white noise into a linear regression will give a p-value of 1.
but you should be able to do a Bland-Altman
Of course you can do one, but it won't be meaningful because it'll indicate a bias where one doesn't exist.
I swear I have seen charts that show the correlation to future wins, or maybe future points, (rest of season? next season? I donâ€™t remember) for DVOA, wins, points, etc, and show DVOA performing the best.
I'm pretty sure they're there, too. In any case, here's a separate one comparing yards per point, DVOA, and yards.
Points is a very, very crappy statistic any way you look at it due to the quantization. That you can see from the preseason article.
91: Are you crazy? White noise in a linear regression will give you an r-sqaured of ZERO. That's the whole point... if you're getting 1 then somebody has an over-paramatized model.
91 said p-value
93 said r-squared
I hope this avoids an argument
...which is the whole point. If DVOA has an R of 0.9 that's impressive... unless white noise gives an 0.89. Context.
91: Are you crazy? White noise in a linear regression will give you an r-sqaured of ZERO.
DavidH is right. I said p-value, not r. Might want to keep the terms straight.
unless white noise gives an 0.89.
Except... it doesn't. It's a linear regression. It'll give zero. As you mentioned before.
To expand on #87:
Ok, this is going to get a messy, but I'll try to keep it a legible as possible. I've been playing with the drive data for too long to come to any conclusions. Hopefully someone else can, though.
Here are the drive breakdown. First will be for drives where the opposing offense starts inside their own 30 yardline (I arbitrarily chose the 30, seemed reasonable). Second will be for drives where the opposing offense starts between their own 30 and mid-field. And third will be for where the opposing offense starts at beyond mid-field.
Drives Starting Inside the 30 Yardline
14 of 22 drives
3 drive of less than 1 yrd
2 FGs (3 missed) & 1 TD
17 of 23 drives
1 drive of less than 1 yrd
1 FG (1 missed)
17 of 24 drives
3 drives of less than 1 yrd
1 FG (1 missed) & 2 TDs
19 of 22 drives
2 drive of less than 1 yrd
0 FGs (1 missed) & 1 TDs
Drives Starting Between the 30 and Mid-field
5 of 22 drives
3 drives of less than 1 yrd
6 of 23 drives
0 drives of less than 1 yrd
2 FGs (1 missed)
5 of 24 drives
1 drives of less than 1 yrd
2 of 22 drives
0 drives of less than 1 yrd
Drives Starting Beyond Mid-Field
3 of 22 drives
1 drive of less than 1 yrd
0 of 23 drives
2 of 24 drives
0 drives of less than 1 yrd
1 FG & 1 TD
1 of 22 drives
0 drives of less than 1 yrd
I'll have to chew on that data a little longer. At first glance I see that:
Chicago's had all but 3 of their opponents drive start inside their own 30.
Chicago's only cause 1 drive of less than 1 yard (same as ATL, CIN has had 4 and SEA has had 7).
Chicago's given up the 2nd most yrd/drive (13.3) when the drive starts beyond their opponent's 30 (ATL give 19.5, CIN gives 11.3, and SEA gives just 8.3).
ATL has 4 INTs, CIN has 3 INTs, CHI has 2 INTs, and SEA has 1 INT.
And looking at the box scores:
Seattle's given up 2 FGs while the game was close (score results in a 1 score difference).
Atlanta's given up 2 FGs while the game was close.
Cincinnati's given up 3 FGs while the game was close (1 on a 4 yrd drive starting at the CIN 17 yrd line.).
Chicago's given up 0 pts while the game was close.
Man. It looks like offenses just hit a complete wall at midfield for Seattle. That's crazy.
You are forgetting fumbles, though, too: Atlanta's opponents have fumbled 3 times on offense, Chicago's 3 times, Cincy's 4 times, and Seattle's 7 times.
Combined turnovers are SEA: 8, ATL: 7, CIN: 7, CHI: 5.
Re: 73, 76, 79, 80, 81, 82, 83, 85, 86, 88, 90, 91, 93, 95, & 96
My head hurts. I hate you both.
Yeah, I didn't feel like going through the play-by-play and the drive chart only shows fumbles recovered by the defense (which FO treats as random).
Oh, and the other thing I forgot: early-season VOA often bounces around a ton due to the importance of turnovers and the small numbers of them.
See Cincinnati, last year.
It'll balance out in a few more games once the sample size gets large enough.
Also, why are we even discussing this? The FOX Rating for defense (which is the best predictor for end of season DVOA, not the current one) ranks BAL #1, CHI #2, and SD #3.
#100: Unfortunately doing that you miss the sack-fest on Kurt Warner last week. Of course, we all know it's just Kurt Warner, so... but still, forcing 4 fumbles in a day? That's a hell of a performance.
2 of Warner's fumbles were unforced and came on aborted plays, according to the play-by-play. That includes one play where Warner fumbled the snap, recovered it, was sacked, and fumbled again.
Where's your editor? ;) The Bears/Lions 23-20 game was the last week of 2000, not 2001. The last week of 2001, the Bears were 12-3, and probably would not have been 7 point dogs to the Lions.
For those of you who are into Scrabble (you know who you are) ... LEINART has valid anagrams
1.) Am I the only person for virtually every post from about 80 onwards is so far over their head they can't even see the bottom of it?
2.) Are Vikings Pirate's? I'd argue no, because they are more of an expantionist culture, moving into areas all over northern Europe, whereas Pirates don't tend to spread their culture, but were more like parasites for a single culture/area. Anyone?
1) I see something that resembles the bottom in the far-off distance, but just bearly.
All I know is that there are two pirate-themed teams in the NFL, but not a single ninja-themed one. That's just plain wrong.
I've only read Wanker79(I apologize, but I had actual work to do for a change) so far, but his quote:
"Chicagoâ€™s given up 0 pts while the game was close."
Hello? Isn't that the point of a Defense? What moere are you looking for?
Turnovers I guess.
Technically, for instance, Baltimore actually allowed -4 points on defense when the game's been in question.
If you really want the executive summary of everything else: look at the FOX Rating for defense. It's #1 BAL, #2 CHI, #3 SD.
The VOA rankings will fluctuate a lot and settle down as the data set grows. The FOX Rating is more predictive right now. Which is why it's used for the rating.
94 ad 96:
I apologize for the mix up, however I don't ever pay attantion to someone who quotes p-values.
You are either significant or you aren't... there is no pro-bowl for p-values.
110 - "The VOA rankings will fluctuate a lot and settle down as the data set grows. The FOX Rating is more predictive right now. Which is why itâ€™s used for the rating."
That's fine, but I expect more from an 'elite' statistic.... adding in more variables may help with your p-value, but this DVOA crap should *explain*, not have to be explained.
If it's the latter you've picked a poor stat son.
like we've said, we're not using DVOA right now. We're using VOA. Theres no opponent adjustment yet, so its just value over average. It just tells how well teams are playing at this point, and doesnt take into account that everyone plays well against the raiders.
Once we hit week 4 or 5 and all the teams are connected, it will start adjusting for poor teams, and Baltimore/SD will start being penalized for playing the absolutely awful raiders.
The sample size is simply too small at this point to derive anything from it.
Thatâ€™s fine, but I expect more from an â€˜eliteâ€™ statisticâ€¦
I would love to see a statistic which can generate significant results from insufficient data. That'd be really impressive.
The 2006 VOA data does not have enough data yet to really be a strong indicator of the season. That's why the FOX ratings exist, as they add in historical data.
I really don't think you're understanding the statistics much at all. The variables weren't tuned to maximize p-values. That's pointless, unless you're going to retest it on an independent data set. They were tuned to maximize correlation.
i'm surprised that the colts are so heavily favored this weekend. if freeney, sanders and harper don't play (they're all questionable for sunday), i think a legitimate case could be made that the colts have the worst defense in the league. but i guess time will tell!
113 and 114: Yeah, I get it once more games have been played things will change.
*However* my point is that without any of those adjudments VOA appears to be a poor indicator.
As I've said, please explain how the Falcons' D bettter than the Bears, which your ratings would indicate
This is a "play by play" stat, so you should be able to tell me what exactly it is that the Falcons are doing tht the Bears aren't.
Is it turnovers or sacks? It certainly isn't yards, so make my pea brain understand the significance of what your rating is telling me.
Iâ€™m a Ravensâ€™ fan, so Iâ€™m not hear to cheer about Chicago
Thank God! I thought the Bears were about to taste the hot fire of the FOMBC.
117: It is true however... I don't give a damn about how good the Bears'D is... I only hope it is worse than the Ravens'... but I never would have thought you guys were approaching "middle of the pack" shit... but that VOA says it, so it must be true.
See'y'all in the Super Bowl.
This is a â€œplay by playâ€? stat, so you should be able to tell me what exactly it is that the Falcons are doing tht the Bears arenâ€™t.
Is it turnovers or sacks?
You still have clearly not read about FOs methodology. Please read about what the numbers mean before continuing this discussion.
119: In fact, I have read about FO's stats... is that your defense on how inaccurate they are? Read more to know their uber-ness?
I agree with Tanier that the standard stats won't capture how good the Bear's D is... however, the FO stat says they are WORSE than what the NFL rates them... why don't you explain that to me since you've already done the readinng.
Is it turnovers or sacks?
I'm pretty sure it's turnovers. See the list above for total turnovers.
It happens pretty much every year. A few teams luck out and get turnovers at high rate early on, and so DVOA on defense thinks they're the Second Coming - and to be honest, so does the mainstream press (see the Bengals, last year) as there's no real way to tell that they're not an insane defense when they're picking up 4 turnovers a game.
Things calm down when the opponent adjustments realize that, for instance, everyone can sack the Raiders and everyone can intercept Chris Simms.
As Iâ€™ve said, please explain how the Falconsâ€™ D bettter than the Bears, which your ratings would indicate
What part of "use the FOX ratings, which are more accurate early in the season, and have the Bears well ahead of the Falcons" didn't you understand?
There's a reason that the rankings on Fox Sports use that rather than the pure VOA rankings.
as long as somebody besides us falcons fans are the new falcons fans, I'll be happy.
122: What part of "explain why your stat sucks" don't you understand? We aren't talking about FOX we are talking about how innacurate DVOA is.
Riddle me this Batman... is Chicago the 7th best D the league?
Riddle me this Batmanâ€¦ is Chicago the 7th best D the league?
Probably not. Has there EVER been a statistic that ranks every single team in the order you think they should be ranked in? How about your favorite stat, points... Was Indianapolis the 2nd best D in the league last year?
125: I'm just asking for an explanation instead or pretendind it didn't happen.
And yes, last time I checked, points win games, not DVOA.
126: Take a look at the play-by-play from last week..
Bears D vs Lions: 7.7 yards per pass, 3.3 yards per rush, 3 fumbles, 13 first downs. Definitely good numbers, but ..
Ravens D vs Raiders: 6 yards per pass, 1.5 yards per rush, 5 fumbles, 3 ints, 9 first downs.
So the reason the Ravens have an out-of-this-world score, and the Bears just have a good one: more fumbles, more INTs, fewer first downs. All of those are huge factors in VOA.
Now, with a little common sense I don't think anyone expects the Ravens to keep up this pace. But it did happen and that's what goes in to the calculation.
Iâ€™m just asking for an explanation instead or pretendind it didnâ€™t happen.
OK, so first of all, since
A) total defensive VOA is basically a weighted average of the passing and rushing VOA
B) the top teams have faced almost twice as many passes as rushes
C) the 6 teams ranked higher than Chicago in total VOA all have better passing VOA
we can agree that explaining why those teams are ranked better against the pass will pretty much explain why Chicago is only ranked 7th, right? If not, this whole thing is pretty pointless.
VOA looks at the results of individual plays, not drives or games or downsets or any other chunk, so let's look at everything on a "per play" basis.
Yards per attempt:
1st downs per attempt:
sacks per attempt:
interceptions per attempt:
So Chicago is 7th, 6th, 7th, 2nd, and 6th in these. If I've forgotten to include something important, let me know. I've intentionally left out scoring because, unlike these 5 measures, scoring is HEAVILY dependent on field position, and Chicago's opponents have had the worst starting field position in the NFL, according to the drive stats (linked in my name).
Let's go team by team. For each team I'll list whether they are better (+) or worse (-) than Chicago in the 5 categories above (same category order):
BAL + + + - +
SDG + + + + +
JAX + + + - +
SEA + + + - -
ATL + + + - +
CIN + - + - +
I think that should pretty much do it. The teams ranked ahead of Chicago have, on a per play basis, simply allowed fewer negative plays (1st downs, completions, yards) and made more positive plays (sacks, interception). This doesn't mean it will continue, or that Chicago is the "7th best defense," or that VOA is the best way to judge football teams. But it should at least explain why Chicago is ranked 7th in the single-play-based stat of VOA.
So, to combine my comments in 87 and 128...
Chicago is allowing first downs at a higher rate than these other teams, AND not getting as many turnovers as most of them. So you would expect them to have allowed more long drives, but somehow they have allowed among the fewest. Is this something that will continue, or is it just random fluctuation? And if it continues, what is it about their defense that allows this to happen?
I killed the thread.
FO's Tom Gower checks in from Chicago with a first-person account of what it's like to cover the NFL draft on the scene.
See All XP | NFL XP | College XP
© Football Outsiders, Inc. // Site powered by Stein-Wein // Partner of USA TODAY Sports Digital Properties