Knockout games don't always go the way form and talent suggest they will. Sometimes they do. That's why, for example, in English soccer you see the team that wins the Premier League (the best team) generally don't win the FA Cup (a knockout tournament). The best team is the team that maintains the highest level of performance over an extended season, knockout competitions are decided by who is best on the day in a series of one-off encounters. Of course, with NFL teams playing such wildly different schedules there isn't really a metric to show who the best team is in any given year. A good example is 2007 where the Patriots were clearly the best team in football, but didn't win the Super Bowl because on one day out of 20 they were not at their best and their opponent played above their average performance level.
(Bolded for effect) That's just it, you are using an opinion, not objective information to determine that. You keep saying that other systems can have similar results; then show me one. One person pointed out Mike Silver who is a hair under 74% so at least that poster is trying. It's not that hard to verify. I'm getting a doctorate in Nursing but had to take 4, 700-level statistics courses.
These two quotes pretty much sum it up. I've been putting my trust into DVOA from FO since I was introduced to it in 2007. Obviously the metrics they are using may or may not be skewed based on the formula and parameters they are using to measure each and every play and the weight each particular play holds against their system. From a common sense point of view, no one is saying that the Jets are hands down better than the Packers. But over 100 games against similar competition, the Jets as a team may win 51% to the Packers 49%. The teams are rated by each team (O, D ST), each play, and a weight against each play, pending situation, down-distance, etc. To use a similar system as an example, look at ESPN's new TQR system. They CLAIM to be using a formula that puts greater emphasis on a 3 yard running play in the 4th quarter of a close game over a 30+ yard pass in the 1st Q of a blowout. However, whatever metric ESPN is using is totally fucking skewed based on common sense approach. What made me completely lose faith in TQR was when McNabb gets a higher rating for throwing for 47 yards and no TDs/INTs over Sanchez when he threw for 200+ yards, 2TDs & 1INT in a close game. Maybe they have the ratings reversed. But from a common sense application against the ESPN statistics, its completely ass-backwards. With DVOA, one could technically make the same claim because the actual formula is unknown and maintained as "intellectual property". However, based on metrics being provided by other advanced statistical entities such as PFF, ESPN, ANFLS, etc. against the results of DVOA, AND the common sense approach, it seems that DVOA's formulae are correct in predictions slightly more than all the other metrics. Based on those results, the logical choice is to trust DVOA over other statistical systems. From a high level, it obvious why people would jump to the conclusion that DVOA is FUBAR cause the Jets are ranked above GB (especially from an opposing team's view). GB is 8-0 and the Jets are struggling to be 5-3. That's the surface. But lets dig a little deeper using our own super-basic metrics.... The Jets ST is bar-none heads and shoulders above any other ST in the league. Lets make ST worth half of what the O & D are because they spend the least amount of time on the field. Our O is mid-pack (17th?). GB O is #1. Our D is top 10 or better. GB's is a sieve. Weighting System: Top 10 = 1 point (ST = .5) 11 ~ 22 = .5 points (ST = .25) Bottom 10 = .25 points (ST = .125) Jets O = .5 Jets D = 1 Jets ST = .5 Total: 2 GB O = 1 GB D = .5 GB ST = .25 Total = 1.75 So we end up with similar results from DVOA where the Jets are ahead of GB even though GB is 8-0. I'm trying to make a point that based on what ScotsJet said about being the best team over the long haul against similar competition. I'm sure that the Jets records would be a lot better if they played the NFC west the last two years, but the weaker competition is probably dragging down GB's overlay points in the DVOA system because the competition they have played up to this point isn't as tough as ours to this point. I agree that half a season does not make for a great sampling size and I'm not about to go researching the last 20 years, but after by brief experience with looking into all these systems over the last few years, its definitely safe to say that of all the advanced metrics systems available, DVOA is the best bet.
I understand that DVOA has some merit, despite the things that confuse me about it. What I am saying is that, since there is a large discrepancy between the rankings of DVOA and other "advanced ranking metrics", can anyone either (a) prove that DVOA has some sort of statistical advantage over these metrics, or (b) provide a logical reason why the Jets are (and should be) ranked so much higher than the other metrics place them. If no one can do either of these things, this particular ranking just seems like an ordinary abnormalty in their system, or the result of an inbalance in their formula. They have admitted to not understanding why their system acts certain ways before. The "advanced ranking systems" that I referred to are as follow: Pro-Football-Reference.com SRS: http://www.pro-football-reference.com/years/2011/ Jeff Sagarin's NFL Rating Index: http://www.usatoday.com/sports/sagarin/nfl11.htm Advanced NFL Stats Team Adjusted Success Rate: http://wp.advancednflstats.com/teamSR.php
Additionally it's funny because you can read the bitter disappointment and confusion in every word of the FO explanation. They hate that their beloved system puts the Jets first.
Edited for length. Fantastic post. @GoPats88, the first and last links are using simple statistics, not situational statistics. The middle link, I do not know the methodology without looking into it more. Situational Statistics are MUCH more important then simple statistics. Which is better: a twelve yard pass on 3rd and 15, or a 6 yard pass on 3rd and 5? The first and last links say a team is better if they complete the 12 yard pass, which is ridicolous because that means you have to punt instead of getting a first down. Go by whatever ranking system you want, and if you get 74% head to head, let us know.
Some of the stuff you listed aren't really complicated metrics, hell one of them is even has SIMPLE in its acronym. Haha, what's an ordinary abnormality. An abnormality isn't ordinary by definition. From what I've read on DVOA, it weights certain plays more highly than others. As the above poster mentioned "Situational Statistics". For example, a first down when the score is close is worth more than a first down for a losing team during a blowout. Since the Jets have played just about everyone close, they're probably accumulating a lot of points in that sense, along with the good special teams, etc. The system very well may have a soft spot for teams like the Jets, the Jets were rated highly in the middle of last season too, but then again the Jets made the AFC championship game, so maybe it knows something? The Jets have also played a pretty tough schedule so far so a lot of the situational stats (first downs in close losses) came against good teams like Baltimore and New England. I wouldn't read too much into it though, from what I remember DVOA becomes more accurate as the seasons goes on.
A quick glance at PFR shows that although GB vs Jets heads up shows that GB's point differential is much greater per game than the Jets, the Strength of schedule is .4 difference. up to this point. As GB's schedule gets tougher (if it does) the number might be higher but that may change the metrics quite a bit. This is especially true after a full season of games, or as opponents become weaker or stronger down the stretch. The number could also get lower if upcoming opponents are dropping like flies and start to fall out. Remember the SoS isn't a constant number. The Raiders were a better team until Campbell went down. Now they are sucking ass. If the Pats were 7-1 or 6-2 coming into this game, the Jets SoS would probably be higher too because of the Pats WL record. The Rams and Jags are at the bottom of the league and thier SOS is 3.5 & 3.2 respectively. So it shows you how much more difficult thier schedules are based on their opponents. As the Giants and SF records increase in Wins, the SoS of the Rams will increase as well. If the Pats or Bills collapse toward the end of the year (hypothetical) I am sure the Jets SoS will drop as well making it seem they played a softer schedule. But the Jets still had one of the toughest schedules in the league. Before Manning went out for neck surgery, the Colts Pats game would have been a tough must see game. Now its knocked off the Sunday night schedule. So everything is in flux until the regular season closes - that's an obvious statement. But the DVOA metrics are ahead of everyone else from a prediction stand point. I think that's the point of the argument - that DVOA represents the best of advanced metrics thus far.
If DVOA was that good, those guys wouldn't be running a website, they'd be raking in cash from Vegas.
I don't get what a band that only had one major hit "Whip It" has to do with football? All these rankings, different states to rate teams and players are bullshit. I realize that's what makes football interesting during the week but every stat/ranking in the world did not have the Rams beating the Saints, but it happened. Being highly ranked or at that top of statistical categories is fine but I've come to the point where all I care about is how many wins do we have. I got too wrapped up in the anti-Jet talk early on and assholes proclaiming the season is over in week 5. It's just like last year, the Seahawks won their division at 7-9, and beat the Saints. I wonder what their DEVO was? (sarcasm)
They're making plenty of cash with that site, I'm sure. Aaron Shatz even has some TV show on ESPN. Besides, ranking the teams using stats is different than playing point spreads and taking weekly situations into account to make gambling picks. For example, the Jets are going to be rated higher in every ranking system than Denver, but the rating systems don't take into account the Jets have basically 2 days of rest between the Pats game and Denver game.
You are looking at this way to simplistically. Of course upsets are going to occur and no one is ever gonna get 100% of the predictions right. No one is suggesting that. DVOA indicates that the Saints are a better team then the Rams. Not that they will win every single meeting with them. You are discounting it because it is not a crystal ball, but nothing is.
What do you mean "that good"? No one says it is a crystal ball, just that it is the best statistical metric to predict head to head matchups.
That's pretty much it. No system could ever take into account upsets (e.g. Seahawks @ Saints 2010 PO). But from a statistical standpoint and a way to rate the teams over the season its pretty good. No one with 1 functioning brain cell would consider any of these rating systems a crystal ball. But it certainly helps kill off the monotony between Sundays and off season and makes for good discussion...
Here is the problem that people that aren't stat geeks have with DVOA and why in some cases it is counterintuitive. You equal weight Green Bay's Offense to every other unit be it O or D or semi to ST in the NFL. Now in Green Bay's case, how do you objectively weigh GB with Rodgers having a statistical HOF season. Meaning Green Bays Offense is so good it renders a weighting system somewhat less than accurate. Their O neutralizes the negative of their D when it significantly outweighs it. This effect is similar to the Jets only that the Jets are not nearly as good on O but their D doesn't have to compensate nearly as much. Where I am going is, DVOA should include some formula that compensates for how far above a standard deviation Green Bay's offense is. They should devise a multiplier somehow. I guarantee that the 2000 Ravens with that historic Defense run into this same problem. **guys I am at Dr office and typed this from phone I have to run. Sorry will finish and edit later. Feel free to pick apart but keeping in mind I am not finished**
Thank you for providing a post that is more then "LOL Numbers R Stupid!" DVOA numbers are (very crudely) how far away from the average you are. DVOA is actually precisely what you suggested, that teams can make up sub par play in one area with exceptional play in another. GB offense is fantastic at 39.3%, but there defense is not. They have let teams crawl back into games (a big no-no in DVOA) and have actually let up a decent amount of points (15th most in the league). It is at 10.7% (remember you subtract defense) and their special teams is at complete average at 0.0%. That means that their total DVOA is 28.6%. Imagine if the Packers had the Jets Offense; they'd be -4.6% and ranked 19th in the league. So yes it proves your point that GB offense is playing out of it's mind, because if it was not even bad, but just a middling one (like the Jets) then they'd be in the bottom half of the league.
Yes, I know what it does and definitely not in the "OMG it's stupid camp". I am saying its flaw is somewhat similar to what Sux said. It rewards teams that are very good in most to all phases of football rather than being otherworldly in just one phase of the game and mediocre to meh in the others. Even though that one area can carry the team to championship. I am suggesting that there be some kind of additional multiplier to GB's offense. They just have that ability to outscore a team to death. As you say if their offense was just a little worse that team isn't undefeated at this point. GB's offense is an historic offense, lets face it. It is magnifying a flaw inherent in DVOA. That is all as there is no rational person obviously that would say the Jets are better than GB.
