My experiment with the (snicker) DVOA Rankings...

[Taco John]

After our discussion last week of Outsider's DVOA rankings, I thought I'd try a little experiment to see just how good these things hold up to scrutiny. So I gave them a test in the only way I know how to truly test them: actual games. I took the chart with me as I made my picks in my pick'em leagues.

For each game I picked, I looked at where the teams were rated in the most current DVOA rankings chart. Figuring how excited everybody seems to be about these rankings, I hoped for an outcome of maybe 70%... Instead I got a miserable 42% of games right picking based on their team ratings:

Their misses:

19th ranked Green Bay lost to 24th Ranked Buffalo

8th ranked Dallas (who are ranked ahead of the 12th ranked Broncos) lost to the 20th ranked Redskins.

#1 ranked Chicago lost to #26th ranked Miami. To their credit, only football nostaligists got this one right. (To my credit, I would have picked with the nostalgists if I weren't running this experiment, and missed my chance to look like a geeniyus).

16th ranked Atlanta lost to 27th ranked Detroit.

The 18th ranked Vikings lost to the 30th ranked San Francisco (dammit. I'd have gotten this one right too).

The 10th Ranked, 2-5 Steelers lost to the 12th Ranked 5-2 Broncos.

And finally, the 6th ranked Patriots lost to the 9th ranked Colts.



And once again... The Ravens, The Patriots, The Steelers and The Chiefs... All teams rated ahead of the Denver Broncos on the DVOA charts... all teams that the Broncos have beaten.

[Kaylore]

I bet Lidderer isn't going to be very happy with that!

[Lidderer]

The best part about this is that you think your small sample size says something.

That aside, it's been pretty effective for me when betting, and while it's more sound and predictive later in the year, I'd still put it up against your average power ranking any given day.

The team dvoa rankings are always a bit sketchy though. The player evaluations are where it's really at, I find.

It does seem bizarre that DVOA always seems to be so down on denver though...until week 10 or so, which: mark your calendar! or don't.

[DomCasual]

You just don't understand the DVOA rankings, simpleton. They aren't meant to determine wins and losses. You act as if they are supposed to somehow magically determine which teams are better than others. Pfft.

They are none of those things.

It takes a far more advanced mind than yours to understand this. And that's the thing about the DVOA rankings - they allow and encourage pretentiousness. Simply put, people that revere them are smarter than people that don't.

Amen, and amen.

[listopencil]

The best part about this is that you think your small sample size says something.

That aside, it's been pretty effective for me when betting, and while it's more sound and predictive later in the year, I'd still put it up against your average power ranking any given day.

The team dvoa rankings are always a bit sketchy though. The player evaluations are where it's really at, I find.

It does seem bizarre that DVOA always seems to be so down on denver though...until week 10 or so, which: mark your calendar! or don't.

So the DVOA is worthless in week 9 but it kicks in during week 10? Or so?

[Taco John]

The best part about this is that you think your small sample size says something.

That aside, it's been pretty effective for me when betting, and while it's more sound and predictive later in the year, I'd still put it up against your average power ranking any given day.

The team dvoa rankings are always a bit sketchy though. The player evaluations are where it's really at, I find.

It does seem bizarre that DVOA always seems to be so down on denver though...until week 10 or so, which: mark your calendar! or don't.


I'm only working with the sample size I've got. Six of 14 games is a pretty poor success rate under any measure. I figure 8 weeks in, the rankings should be pretty well shaped out, with enough data to provide something useful here. That clearly isn't the case. The DVOA is a great tool for writing articles that Fox will pay for.

[Lidderer]

I think 'pretentiousness' is reserved for Brakhage films and bouqets of tannins, not correalations.

This is probably the same reaction people had to VORP though, and admittedly these things take time to perfect and catch-on in the popular imagination. It's probably not too far off when DVOA or some equivalent gets added to stat-lines much the same way OPS has been bandied about in baseball in recent years.

[Taco John]

You just don't understand the DVOA rankings, simpleton. They aren't meant to determine wins and losses. You act as if they are supposed to somehow magically determine which teams are better than others. Pfft.

They are none of those things.

It takes a far more advanced mind than yours to understand this. And that's the thing about the DVOA rankings - they allow and encourage pretentiousness. Simply put, people that revere them are smarter than people that don't.

Amen, and amen.


You know what the DVOA rankings don't consider? How much commercial time a player gets. If it included that, they would understand just how great Chad Johnson really is.

/end side joke.

[Taco John]

I think 'pretentiousness' is reserved for Brakhage films and bouqets of tannins, not correalations.

This is probably the same reaction people had to VORP though, and admittedly these things take time to perfect and catch-on in the popular imagination. It's probably not too far off when DVOA or some equivalent gets added to stat-lines much the same way OPS has been bandied about in baseball in recent years.


For what purpose?

[Rock Chalk]

For what purpose?

Masturbation.

[Lidderer]

So the DVOA is worthless in week 9 but it kicks in during week 10? Or so?

Not hardly, it's just much more effective. It's weighted DVOA, which means it takes into account your opponents, who would be hard to accurately gauge unless you had a good idea of how they were(say, 8 games). It's still quite accurate at this point, as I'm sure you'll find if you just, i dunno, save their power rankings right now and view them at the end of the year.

[Rock Chalk]

Not hardly, it's just much more effective. It's weighted DVOA, which means it takes into account your opponents, who would be hard to accurately gauge unless you had a good idea of how they were(say, 8 games). It's still quite accurate at this point, as I'm sure you'll find if you just, i dunno, save their power rankings right now and view them at the end of the year.

So, at the end of the year you think Denver will be the 12th best team and Pittsburgh will be the 10th best team? :thumbsup:

Because if you are talking about their end of year power rankings, I can throw something out there that same week which will be a pretty good ranking of the teams in the league after 16 weeks AND I will only use two stats. Wins and Losses as my measuring stick.

As it is, I can look at the teams, gauge how they played WITHOUT making kneejerk reactions to single games and make a very accurate power ranking of the league weekly that doesnt have the Colts at the 6th best team and Denver the 12th best team while ****ty Pittsburgh is anywhere above the 20th best team and thats being generous seeing as how they have exactly one win more than the worst team in the league.

[Lidderer]

For what purpose?

Well I mean how did we evaluate a ballplayer a decade ago even? We tended to look at RBI(practically laughable as a success measure in the contemporary view), or Batting AVG, right? But now OPS is there for the purpose of better valuing a player's overall worth, ie how likely he is to produce runs/prevent outs, etc.

So I'd imagine DPAR/DVOA would serve the same purpose, which see our previous talk about gradkowski's overhyped early numbers.

[listopencil]

Not hardly, it's just much more effective. It's weighted DVOA, which means it takes into account your opponents, who would be hard to accurately gauge unless you had a good idea of how they were(say, 8 games). It's still quite accurate at this point, as I'm sure you'll find if you just, i dunno, save their power rankings right now and view them at the end of the year.


Now that eight games have been played then then rankings should be very accurate. Why is eight the magic number? I saw them after seven and they were ridiculous for the Broncos. The Steelers were the third team (IIRC) that were rated above DEN that we have beaten now. If the ranking system is only accurate after the games have been played it is worthless. If it takes all year to get accurate results that would also make the ranking system worthless. Care to post the results to date?

[Taco John]

What do DVOA team rankings and Nostradamus have in common?

Neither make any sense, and are accurate only after it's too late to do anything about it.

[Lidderer]

So, at the end of the year you think Denver will be the 12th best team and Pittsburgh will be the 10th best team? :thumbsup:

Because if you are talking about their end of year power rankings, I can throw something out there that same week which will be a pretty good ranking of the teams in the league after 16 weeks AND I will only use two stats. Wins and Losses as my measuring stick.

As it is, I can look at the teams, gauge how they played WITHOUT making kneejerk reactions to single games and make a very accurate power ranking of the league weekly that doesnt have the Colts at the 6th best team and Denver the 12th best team while ****ty Pittsburgh is anywhere above the 20th best team and thats being generous seeing as how they have exactly one win more than the worst team in the league.

I don't take it as manna from the football heavens, and they don't either(how refreshing). But it's a good tool for measuring a team's worth and overall performance. I mean look at the 1998 Arizona Cardinals. Mediocre team on almost every level(that includes a 2nd year Plummer), but manages to make the playoffs and defeats a wonky, over-the-hill cowboy squad in a playoff game. But the team was garbage--I know, I watched every single god-blessed game--and lucked out way more than I've ever seen a team do in one year. Well DVOA would have told you that they weren't a legit threat to do anything much THAT year, or the next. I think they went 5-11 the next year with the same nucleus.

And no, I'm not talking about their end of the year power rankings, I mean compare what they are saying now, at this moment, to what transpires.

There's a lot they can't account for yet though, and it's still in its infancy, so while I laud the idea and much of its results, I can also see its limitations and inevitable misfires.

[Lidderer]

man, they better send me a t-shirt or something

[Taco John]

man, they better send me a t-shirt or something


:rofl:

[listopencil]

I don't take it as manna from the football heavens, and they don't either(how refreshing). But it's a good tool for measuring a team's worth and overall performance. I mean look at the 1998 Arizona Cardinals. Mediocre team on almost every level(that includes a 2nd year Plummer), but manages to make the playoffs and defeats a wonky, over-the-hill cowboy squad in a playoff game. But the team was garbage--I know, I watched every single god-blessed game--and lucked out way more than I've ever seen a team do in one year. Well DVOA would have told you that they weren't a legit threat to do anything much THAT year, or the next. I think they went 5-11 the next year with the same nucleus.

And no, I'm not talking about their end of the year power rankings, I mean compare what they are saying now, at this moment, to what transpires.

There's a lot they can't account for yet though, and it's still in its infancy, so while I laud the idea and much of its results, I can also see its limitations and inevitable misfires.



...and yet that Cardinal team won a playoff game. How many franchises right now-this year-would take that result as a positive for the season? Who beat them in the playoffs that year?

[Lidderer]

Now that eight games have been played then then rankings should be very accurate. Why is eight the magic number? I saw them after seven and they were ridiculous for the Broncos. The Steelers were the third team (IIRC) that were rated above DEN that we have beaten now. If the ranking system is only accurate after the games have been played it is worthless. If it takes all year to get accurate results that would also make the ranking system worthless. Care to post the results to date?

I'm pretty sure they come out on wednesday or thursday. Just go to their site sporadically.

And let's be honest, football is a tough game to predict what with countless variables, and that's why their goal is gonna be awful hard to reach seeing as baseball can give you a clearer view of one player's worth much more accurately using just numbers.

If any of us could be picking games at a 58%+ clip we'd be living in Vegas getting heckled by Slap. Finding, or at least searching for, a key to solving games, be it for gambling or other purposes, is better than relying purely on your gut or a coin-flip.

[Lidderer]

...and yet that Cardinal team won a playoff game. How many franchises right now-this year-would take that result as a positive for the season? Who beat them in the playoffs that year?

Was jacksonville better than denver when they beat us in the playoffs years ago? And I don't mean in the banal sports platitude sense of "the best team on the field that day" or whatever. Were they REALLY the better team? No, of course not, and that stuff always has and always will happen in football or any other sport on certain occassions. But it happens a lot less than the alternative, and if you can have a better understanding of just what the alternative is, well that's something.

Minnesota Vikings beat them in the playoffs that year.

[listopencil]

Was jacksonville better than denver when they beat us in the playoffs years ago? And I don't mean in the banal sports platitude sense of "the best team on the field that day" or whatever. Were they REALLY the better team? No, of course not, and that stuff always has and always will happen in football or any other sport on certain occassions. But it happens a lot less than the alternative, and if you can have a better understanding of just what the alternative is, well that's something.

Minnesota Vikings beat them in the playoffs that year.


JAX exposed our lack of a pass rush. Just like teams are doing to us now. They were the better team. BTW, the Vikings were a pretty damn good team that year.

[watermock]

I don't really know how it's worth much if it finally sorts itself out after 16 weeks, if ever. Not much use at that point now is it. It should be sorted out by midseason. They will look at Pitt's 430 yards and probably move them up.

[Lidderer]

JAX exposed our lack of a pass rush. Just like teams are doing to us now. They were the better team. BTW, the Vikings were a pretty damn good team that year.


Well, ya, minnesota was great that year. Better than atlanta even, no?

[Lidderer]

They will look at Pitt's 430 yards and probably move them up.

not at all. Much the same way Plummer had a negative DVOA when he tossed up 499 vs atlanta 2 years back.

[listopencil]

Well, ya, minnesota was great that year. Better than atlanta even, no?


As it turned out, no.

[ZachKC]

man, they better send me a t-shirt or something

:rofl:

[Lidderer]

As it turned out, no.

Little thought experiment. Let's say you are go to New York for 3 days to see the colorado rockies vs the yankees in a 3 game interleague series. Let's also say that you are gonna bet on 2 of the games. Let's further add that the yankees record at that point is 79-30, and colorado is 19-109. So suppose you just watch the first one, to, ya know, see who's the better team, and colorado ends up winning 5-4 or whatever. You're telling me that the next day you're gonna bet on the rockies?

[ZachKC]

Don't bet on the Royals.

[Taco John]

Little thought experiment. Let's say you are go to New York for 3 days to see the colorado rockies vs the yankees in a 3 game interleague series. Let's also say that you are gonna bet on 2 of the games. Let's further add that the yankees record at that point is 79-30, and colorado is 19-109. So suppose you just watch the first one, to, ya know, see who's the better team, and KC ends up winning 5-4 or whatever. You're telling me that the next day you're gonna bet on the royals?


You've got a great point... One which would make a world of sense if we played three game divisional series in the NFL. When we start to do that (it's only a matter of time), I'll get my DVOA shirt and wear it proudly.

[Clockwork Orange]

Don't bet on the Royals.

Or the Rockies.

[ZachKC]

Yes, if the DVOA tells us anything it is that you shouldn't bet on the Royals are Rockies.

Thanks a lot.

[Lidderer]

You've got a great point... One which would make a world of sense if we played three game divisional series in the NFL.

well ya, it's much tougher in football given the small sample size. But it's still advantageous to have some concrete idea of which team is better beyond mere record. And while I did mention only the team's record in the baseball analogy, 100+ games has a way of sorting that stuff out for us. With football we have to go beyond that.

I mean does anyone really think the jets are a .500 team? Philly is only .500 as well? Put philly vs nyj together and i'd go odds on the eagles, even if it's just one game. So would you. And why is that? We might both mention Mcnabb, scheduling, things like that, and DVOA does the same sort of thing, only more precisely and closely-examined. And, unfortunately, with numbers.

[Lidderer]

Yes, if the DVOA tells us anything it is that you shouldn't bet on the Royals or Rockies.

Thanks a lot.

don't shoot the defense team. Listopencil brought up an inane point and I just showed the error in it.

[listopencil]

Little thought experiment. Let's say you are go to New York for 3 days to see the colorado rockies vs the yankees in a 3 game interleague series. Let's also say that you are gonna bet on 2 of the games. Let's further add that the yankees record at that point is 79-30, and colorado is 19-109. So suppose you just watch the first one, to, ya know, see who's the better team, and colorado ends up winning 5-4 or whatever. You're telling me that the next day you're gonna bet on the royals?

I would probably bet on Colorado to win again. Before the game, if I was a betting man, I probably would have bet against the Rockies assuming that they were the weaker team. I really don't get into gambling though and I rarely ever do it. What in the world do the Royals have to do with it?

[watermock]

You've got a great point... One which would make a world of sense if we played three game divisional series in the NFL. When we start to do that (it's only a matter of time), I'll get my DVOA shirt and wear it proudly.

That will never happen. Besides the fact it weights the team with two home games, there are 32 teams. There are plenty of opponents out there.

[Lidderer]

I would probably bet on Colorado to win again. Before the game, if I was a betting man, I probably would have bet against the Rockies assuming that they were the weaker team. I really don't get into gambling though and I rarely ever do it. What in the world do the Royals have to do with it?

ha, ya my mistake. Meant 'rockies'.

But thanks for the honest answer.

[ZachKC]

http://www.baseballballparks.com/royals/images/Royals%202.jpg

[listopencil]

ha, ya my mistake. Meant 'rockies'.

But thanks for the honest answer.

Oh, OK. I was lost. I don't gamble at all. I'm looking at DVOA's ability to determine what the better team is rather than picking a team to bet on.

[Lidderer]

Oh, OK. I was lost. I don't gamble at all. I'm looking at DVOA's ability to determine what the better team is rather than picking a team to bet on.

Well think of 'gamble' as synonymous with determining the better team. It's basically the same thing.

[listopencil]

Well think of 'gamble' as synonymous with determining the better team. It's basically the same thing.


Then I wouldn't use that DVOA system at all. Every time I check it the results are incredibly goofy. I remember TJ pointing out that at one time it was manually adjusted because one of the teams (Indy?) had a ranking that was just incredibly unrealistic. That's a huge red flag.

[DB84FAN]

ha, ya my mistake. Meant 'rockies'.

But thanks for the honest answer.

next time even for example pls do not bring up baseball... i dont think it is a sport even for giving examples..... :yayaya:

Hilarious!

[Blart]

The guy who runs the DVOA rankings reads the mane. He mentioned it several time last year when we flamed him.

[Garcia Bronco]

You can play the probabilities, you can create all the craziest of statistical compulations....but in the end...there is no predictor...and it boils down to luck.

[DarkHorse30]

who was likely told that baseball is for girls......now he brings his rotoworld hack numbers to football.....and get's completely exposed in his first year, ESPECIALLY concerning last year's Bronco team.

Pretty much took him off my list of football hacks to pay attention to. Next.

[Mediator12]

For the record, this does not correlate with W-L's every week but is a measure of success play by play. It can not predict on a game by game basis what will happen, but is designed to see what teams should have success over time. Football can sometimes boil down to luck in a game that averages 130 Plays or so per game. Which play is the most important: the 80 yard run that results in a FG or the 3 yard run that results in the first down. The answer is it depends on the situation.

You can never take the context out of a football game and that is what DVOA attempts to take into account. It says that in any given situation team A is more likely to succeed in situation A than say team B. And it ranks the teams, Offense, Defense, QB's, RB's, OL, WR's, TE's, DL's in ways that can be measured per situation over time. Everyone understands that it only takes one player to screw up any given play and that all 11 have to function properly to succeed the majority of the time. It shows which teams are the most consistent executors in any given situation and nothing more.

However, like anything else it can not predict who is going to screw up in crunch time with the game on the line. It will not predict that Brady would have 4 INT's that gave the Colts defense the stops it needed to keep them under 30 points. It can not predict that PIT would turn the ball over nine times at home on consecutive weeks and lose to a team with barely a hundred yards of offense and no scoring drives that were not off TO's. That is why everyone tunes in each week to watch the games, because you never know what is going to happen from week to week and matchup to matchup.

RANT MODE ON

And the smarter than you comments are absurd. If you wanted to take the time to comprehend and had an open mind anyone could easily understand the premise of this theory. It is not just for smart people, but people who want to see what is happening beyond poor public perception. The problem is people are too lazy to take the time to actually study what this means and would rather assume what they see is not emotionally attached to their point of view.

For some reason, people absolutely love to slander what they do not understand. It is a core human concept proven over history and is the basis for hatred and social racism. I understand that people do not like the way certain things look, but why spend the negative energy to attack something ??? What does that do for people?

My problem is people prejudge that it is bogus and are making up arguments that do not even apply to what DVOA and DPAR are meant to do. You are trying to disprove something before you take the time to understand what it is. And, you will not look at what it is actually measuring.

The mere concept of DPAR is way beyond what people here want to believe. It tries to measure performance on a game by game basis, instead of lumping history and personal favor into the mix. It takes away bias and rates how well that players performance affected the outcome for that team in that game. It could care less how they did last week or will do next week. It focuses on overall performance not perception of the player. Then, the overall DPAR number's are how well that person has performed over the course of those performances to date.

What it can not do is see how that player's presence affected the play around them by getting double teamed, having coverage rolled their way, having eight men in the box, missed tackles and the like. It has severe limitations at this point on a game by game basis, but over time it levels out the field and shows why teams were competitive or not and WHY. This is clearly explained in the section on their site that goes over what these stats are designed to do. Simply from reading the numerous comments about DVOA and DPAR on this site, I would bet not one of those people have read that part but are certainly willing to criticize it and jump on the anti-bandwagon.

[epicSocialism4tw]

Statistics 101: Learn to read statistics.

[Garcia Bronco]

Regardless, any measuring stick has conceptual criteria which may or may not describe the actual criteria. So there is bias, confound, and subjective aspects in any battery, archival analysis, or scale.

[Lev Vyvanse]

I like these rankings much better.
http://www.dolphinsim.com/ratings/nfl/

[Lev Vyvanse]

I hate the DVOA but Taco you didn’t account for home field advantage.

[Merlin]

Shatz is on record for stating that he HATES the broncos because they constantly beat his favorite team (the Paties). So of course, he sought to develop a system that would seem to be objective, but it was designed to denigrate the Broncos (just see how poorly it did for the first 10 games of last yr, and how poorly it has done this yr).



Before anyone goes half flaked, only half of my statement was not in jest, he truly does hate the Broncos.
For the record, this does not correlate with W-L's every week but is a measure of success play by play.

I have been following outsiders for a while for a couple of reasons. One, they have pretty good stats. Two, it annoys the hell out of me that they are so incompetent at gauging the Broncos. As Liderer may remember, my assumption was that they were placing too much weight in certain categories that inadvertently penalized the Broncos. However, no I think their problem is more biased then that. THIS IS A SYSTEM FOR MEASURING HOW PRETTY YOU WIN. That is, because they try to measure each and every performance and don't (can't?) consider the overall goal of the scheme (e.g., bend and don't' break defences, not running up the score, etc.), they easily penalize teams that play with a strategy that is more global in nature.

This is a system that would work great in college football where points for style count, but in pro football, it has limitations (but then all statistical analysis by its very nature will have some limitations). Shatz has stated that his system is thrown off when one team spanks another team (e.g., KC-SF, and Pitt-KC), and gives them far more points than they deserve. Conversely (and this is what he totally misses). Teams that choose to go conservative on offence and/or use bend don't break D's will be penalized in each game, and will be heavily penalized when the strategy is used against teams that should be creamed.

The above notwithstanding, once he has enough data in his system (at least 11 games played by each team) his system seems to compensate for its bias and is able to make better judgments about each team.

PS I doubt the system could ever be used to predict winners and losers because of some of the factors mentioned above (e.g it would difficulty addressing changing performance of individual athletes because of injuries or changes in strategies, or how strategies may work against one team but not another> Denver plays SD better than Indy plays SD, and Indy...).

[Lidderer]

just another little example here of why DVOA is better than your run-o'-the-mill power ranking commentary.

This week's subjective Fox power rankings says this about the falcons:
"The Falcons suffered a mid-season collapse last year that kept them out of the playoffs. If they want to avoid the same thing in '06, they can't be losing to an undermanned Detroit team."

Now on the surface that sounds sensical and fine, if not a bit too obvious of an observation. But it doesn't really say anything beyond a sports cliche("mid-season collapse"). Why did they collapse? Did they stop playing as well as they had earlier in the season? I mean, what?

DVOA had atlanta ranked around 18th even while they were at 6-2. This suggested the team was overvalued. They didn't 'collapse' in the 2nd half of last year, they simply played how they'd been playing and it finally caught up with them.

If they want to avoid the same thing in '06, they have to be better than they've been thus far. That's an important distinction, and one that DVOA allows us to see.

[listopencil]

That's an important distinction, and one that DVOA allows us to see.


And yet this year they had three teams ranked higher than us that we beat. Including the now 2-6 Steelers. I doubt that PITT is going to pull off some sort of mid-season explosion that pushes them to the top. Even if they were to win out they would be 10-6 and most likely missing the playoffs. I do agree that the Falcons were overvalued just because I think Vick was. But that's from watching him play.

[Lidderer]

And yet this year they had three teams ranked higher than us that we beat. Including the now 2-6 Steelers. I doubt that PITT is going to pull off some sort of mid-season explosion that pushes them to the top. Even if they were to win out they would be 10-6 and most likely missing the playoffs. I do agree that the Falcons were overvalued just because I think Vick was. But that's from watching him play.

Well ya, if every team that was ranked higher beat every lower ranked team than the nfl would stop playing games after week 6 and just ask FO to simulate the season. Hell, vegas would demand it. You're expecting it to be perfect, which is silly. Even a 58% correct rate is incredible. You want 100%.

At any rate, the point I was making in the last post was more about what's different between your average commentary and DVOA. That point still holds, and you didn't address it for good reason.

[Lev Vyvanse]

I doubt that PITT is going to pull off some sort of mid-season explosion that pushes them to the top. Even if they were to win out they would be 10-6 and most likely missing the playoffs.

You are still missing the point of the DVOA ratings. Each week they are compiled to predict who will win in the future. Not to see who will have the best record at the end of the year.

[Lidderer]

You are still missing the point of the DVOA ratings. Each week they are compiled to predict who will win in the future. Not to see who will have the best record at the end of the year.

Mediator(or Herc, I get them confused for some reason) clarified much better than I, but it's less about predicting and more about giving you a better understanding of, i dunno, let's call it 'deserved wins' or something.

Which, in turn, should assist in predicting.

[listopencil]

Well ya, if every team that was ranked higher beat every lower ranked team than the nfl would stop playing games after week 6 and just ask FO to simulate the season. Hell, vegas would demand it. You're expecting it to be perfect, which is silly. Even a 58% correct rate is incredible. You want 100%.

At any rate, the point I was making in the last post was more about what's different between your average commentary and DVOA. That point still holds, and you didn't address it for good reason.

Are you referring to the ATL point? That's just the other side of the same coin and illustrates what is wrong with all power rankings. I thought you would get the connection.

[listopencil]

You are still missing the point of the DVOA ratings. Each week they are compiled to predict who will win in the future. Not to see who will have the best record at the end of the year.

I must be. It seems worthless to me.

[Lev Vyvanse]

It seems worthless to me.

You’ll get no argument from me.

[Rock Chalk]

So Pittsburgh "deserved" to win more than their 2 games this season but havent.

And Denver hasnt "deserved" to win their 6 games this season but they have.

I get it now.

Seriously DVOA is worse than kneejerk power rankings. Knee jerk power rankings are opinions on who is the best team NOW. CNN, CbS, and ESPN and Fox's subjective rankings all base on who is playing well now. And its not ALWAYS about win/loss as all but CBS have Denver at #2 and Chicago at #3 or lower on one of them.

DVOA doesnt take into account for how teams are overcoming their shortcomings and winning games, which, is the ultimate measure of how good a team is at any given moment.

Your case on Atlanta last year at 6-2 but being 18th in DVOA was wrong. They, 8 games into the season, were not the 18th best team in the league. They were among the top 10 easily. AT THAT TIME. Later, they collapsed and whileDVOA is OK (note; OK, not great, not good, but OK) at predicting future wins, here's something to chew on: DVOA didnt have one team predicted to win more than 12 games last year. Not one.

Indy, Denver, Seattle all won more than 12 games and I believe three other teams won at least 12 games.

So much for their "predicting" future wins theory.

DVOA is flawed in that it applies (rightfully so) weights to certain categories. However right it is to weight certain categories, that is also what makes it flawed insomuch that who decides what's more important? DVOA doesnt take into account idiotic motorcycle accidents or appendectomies, it JUST judges plays or blown ACLs or rare drops by sure handed receivers or the lack of fumbles by fumble prone running backs or any other of about a bazillion variables that go into football PLAYS.

Personally, I like DVOA because it gives me something to poke fun at and exploit holes in. Probably much the same way hackers like Windows. But If I want a real OPINION on who is who in the NFL on a weekly basis, I will stick to the subjective power rankings and their so-called "kneejerk" reactions to wins and losses each week. At least they generally have the good teams at the top in any givenweek and the bad teams at the bottom with the average teams in the middle.

[Lidderer]

Your case on Atlanta last year at 6-2 but being 18th in DVOA was wrong. They, 8 games into the season, were not the 18th best team in the league. They were among the top 10 easily.

Why? How? What caused this 'collapse'?

[Rock Chalk]

I don't take it as manna from the football heavens, and they don't either(how refreshing). But it's a good tool for measuring a team's worth and overall performance. I mean look at the 1998 Arizona Cardinals. Mediocre team on almost every level(that includes a 2nd year Plummer), but manages to make the playoffs and defeats a wonky, over-the-hill cowboy squad in a playoff game. But the team was garbage--I know, I watched every single god-blessed game--and lucked out way more than I've ever seen a team do in one year. Well DVOA would have told you that they weren't a legit threat to do anything much THAT year, or the next. I think they went 5-11 the next year with the same nucleus.

I wanted to come back to that. Specifically the Arizona part. DVOA said Arizona wasnt much of a threat to do anything eh. Gee, did ANYONE think they were that year? As good as they played no one thought they were much of a thread and NO ONE (not even DVOA) thought they would beat the Cowboys. Yet, they did.

Playing the games: 1
Everyone Else: 0

As for lucking out in a single year, THATS THE BIGGEST FLAW IN DVOA. That random attribute of luck that there is no number, weight or way to account for. As luck would have it, Ben is a ****ing idiot and Jake Plummer had two picks dropped. Who knew? At the end of the day, the best gauge of who is better is wins and losses and all the other stats are meaningless. By week 11, even Shambibi al Zulu Shakka in Nigeria knows who the good teams in the NFL are, yet Aaron needs 11 weeks of data to determine that with fantastic forumalae. Ha!

And no, I'm not talking about their end of the year power rankings, I mean compare what they are saying now, at this moment, to what transpires.

There's a lot they can't account for yet though, and it's still in its infancy, so while I laud the idea and much of its results, I can also see its limitations and inevitable misfires.

Yeah, and as of last week they had Pittsburgh ranked in the top 10 expecting them to improve. If you can see its limitations and misfires then you MUST understand that DVOA !> Kneejerk Rankings.

[Rock Chalk]

Why? How? What caused this 'collapse'?

Why: 6-2
How: By winning 6 games.
What Caused This Collapse: Id have to go back and watch the games Atlanta played AFTER the midpoint of the season.

My contention was that DVOA at the midpoint had Atlanta at 18th and, AT THE MIDPOINT they were not 18th in the league.

They may have been 18th at the end of the year, but that MAY have been explained by DVOA then again, it could be just that random bit of LUCK that DVOA nor anyone else can account for. One bad call here, one crappy coaching decision there, one bad throw by Vick here. Who knows why. DVOA got lucky once, hell Bob gets lucky from timeto time too.

[Lidderer]

let's look at atlanta last year.

Win over jets: vinny fumbles 3 times(dvoa accounts for this!), 17 points came from these turnovers in jets territory, and fumbles are a matter of mostly luck in terms of retrieval; vick has his lowest qb rating of his career. Falcons still win.

Win over NO: vick puts up 106 passing yards; phantom flag on the last-second game-winning Saints FG; Falcons outgained by 200+ yards. This was the new orleans saints also, remember.

other teams they beat: buffalo, miami, minnesota pre-brad johnson.

Top 10 my ass. They didn't collapse, they just played the same way they'd been playing and stopped being fortunate.

[Lidderer]

Why: 6-2
How: By winning 6 games.



You're better than this.

[Rock Chalk]

You're better than this.

OK, fair enough.

Let's looka t Denver last year.

I can remember at LEAST 10 near picks by opposing defenses during Plummers no pick streak.

DVOA accont for that? We were "lucky", because luck is a dominating factor in any sporting event.

So Atlanta was lucky, thats how. Still, better lucky than good if it gets you 6 wins in 8 games? At that point in time, Atlanta was not the 18th best team in the league. Truth be told at the end of the year they were note ven that low. Certainly in the teens but not the 18th best team in the league.

Fumbles I understand are measured in some mystical luck formula by Aaron in which he gets to determine how lucky recovery is.

But is it lucky to get the recovery when receiver fumbles the ball in the secondary and 6 opponents are near the ball while only said fumbler for fumbling team is near the ball? No sir. Thats not luck. Its luck if you DO recover it as the fumbling team.

[Lidderer]

I can remember at LEAST 10 near picks by opposing defenses during Plummers no pick streak.

DVOA accont for that? We were "lucky", because luck is a dominating factor in any sporting event.

.

I can also remember 10 near picks missed by bailey and co. That stuff tends to even itself out, or come awfully close to it.

Luck will always be a factor, of course. They are trying to be perfect, knowing full well that the feat is impossible. But getting closer is good.

There will always be things we can't account for, especially in football. It doesn't mean we can't account for some things though.