Football Outsiders Team DVOA

[ivotuk]

Check out #7

 

[ivotuk]

Also, I love reading Ben Muth's stuff in "Word of Muth" He's a former offensive lineman, and knows his stuff


Ben Muth -- Word Of Muth

Not Just Dak—Cowboys Struggled in Run Defense Too

Not Just Dak—Cowboys Struggled in Run Defense Too

Offensive line expert Ben Muth was impressed with the Tampa Bay Buccaneers' run blocking on Sunday night, but the Dallas Cowboys sometimes left him scratching his head.

www.footballoutsiders.com

 

[Welshers]

Check out #7

There are some serious issues with this list... Dallas #3?? Tampa and Jacksonville 8 and 9? How in the world does dova work

 

[toffee]

There are some serious issues with this list... Dallas #3?? Tampa and Jacksonville 8 and 9? How in the world does dova work

Fans are supposed to follow, not to challenge expert opinions, and most do follow.

 

[onanygivensunday]

There are some serious issues with this list... Dallas #3?? Tampa and Jacksonville 8 and 9? How in the world does dova work

From 30,000 feet, DVOA is an index developed by Football Outsiders of how effective a team's defense and offense perform against their prior opponents given all the down and distance situations they have faced.

A team scores well if they get a 1st down facing 3rd and 10... but doesn't score that well if they fall a yard short of a 1st down.

There is a separate Defensive DVOA score... and one for offense as well. The list here is for a composite index... the average of offense and defense.

Later in the season, FO also calculates a Weighed DVOA that values the last several games over the ones earlier in the season.

 

[DarkVictory23]

There are some serious issues with this list... Dallas #3?? Tampa and Jacksonville 8 and 9? How in the world does dova work

Fans are supposed to follow, not to challenge expert opinions, and most do follow.

This isn't really a fair comment on DVOA rankings. They aren't opinion, it's a formula.

Rather than being based on traditional metrics, it measures things on a per-play basis. It rates teams based on how successful they are per play compared to average, adjust it by the team they're facing, and then weight it based on how recent it was.

In the case of Dallas and Tampa Bay, both teams are being carried by their defense (and in Dallas's case, special teams) while both have below average offenses.

It's not a perfect system (none are!), and I've gone into issues I have with it before, but I do find it a more interesting metric than traditional power rankings.

 

[ivotuk]

There are some serious issues with this list... Dallas #3?? Tampa and Jacksonville 8 and 9? How in the world does dova work

You didn't read it did you? Look at the numbers, total DVOA...

You can't get much lower than 5.6% (Jaguars). What this tells us is that Buffalo is really really good, but most teams at this point in the season are "meh."

 

[Lagartixa]

The Seahawks are seventh in overall DVOA. Breaking that down, they are...
Third in offensive DVOA,
19th in defensive DVOA, and
20th in special-teams DVOA.

DVOA appears to be a good measure of productivity (and DYAR a good measure of production). My biggest problem with FO's stats is that the folks at FO think they have to protect the formulas used to calculate these things as trade secrets, so the stats end up being black boxes. They do say what goes into the calculations and roughly how each factor affects a team's DVOA rating, but they don't give explicit formulas.

I'm guessing somebody with a lot of free time to waste could probably get reasonably close to reverse engineering DVOA and DYAR, possibly using machine-learning techniques, but I don't think it would be worth the time it would take, except maybe to somebody looking for a learning project and/or a portfolio project for applying for jobs.

 

[FPD]

The Seahawks are seventh in overall DVOA. Breaking that down, they are...
Third in offensive DVOA,
19th in defensive DVOA, and
20th in special-teams DVOA.

DVOA appears to be a good measure of productivity (and DYAR a good measure of production). My biggest problem with FO's stats is that the folks at FO think they have to protect the formulas used to calculate these things as trade secrets, so the stats end up being black boxes. They do say what goes into the calculations and roughly how each factor affects a team's DVOA rating, but they don't give explicit formulas.

I'm guessing somebody with a lot of free time to waste could probably get reasonably close to reverse engineering DVOA and DYAR, possibly using machine-learning techniques, but I don't think it would be worth the time it would take, except maybe to somebody looking for a learning project and/or a portfolio project for applying for jobs.

I don't blame them for protecting their formula. People in the food industry do it all the time. But seriously, as a former mathematician, accountant, and businessman, I would definitely protect my company's proprietary mathematical formula. Especially if it were profitable to do so. After all, businesses are in business to make money.

 

[DarkVictory23]

Just want to add that if you're looking for a Footballoutsiders equivalent to a Power Ranking, I wouldn't go with DVOA but rather their Super Bowl odds.

That does consider W-L record and more closely relates to what people think of when they think of Power Rankings.

If you look at those, TB is still high, but Dallas drops below KC and Jacksonville goes to bottom 10. Seahawks are #9.

 

[Welshers]

You didn't read it did you? Look at the numbers, total DVOA...

You can't get much lower than 5.6% (Jaguars). What this tells us is that Buffalo is really really good, but most teams at this point in the season are "meh."

Of course I read it. I just happen to have no understanding of what these numbers represent. I mean you can't get much lower than 5.6% but yet at the same time there's over 20 teams lower?

 

[DarkVictory23]

Of course I read it. I just happen to have no understanding of what these numbers represent. I mean you can't get much lower than 5.6% but yet at the same time there's over 20 teams lower?

Think of it terms of over average on an average play. On any given play, the Jaguars are expected to be 5.6% more successful than an average team. That's not much above average. The teams below them have negative DVOA, meaning on an average play, they do worse than average.

Seahawks are performing 11.4% better on the average play, but that's not evenly divided. We do much, much better on our average offensive play but worse than average on our average defensive play. (Though Seahawks actually had the best defensive DVOA in the NFL over the last two weeks.)

 

[onanygivensunday]

And iirc, what may be a bit confusing is the better defensive teams have negative Defensive DVOAs... but total DVOAs calculations don't consider the negative defensive DVOAs as negative numbers.

 

[Welshers]

Think of it terms of over average on an average play. On any given play, the Jaguars are expected to be 5.6% more successful than an average team. That's not much above average. The teams below them have negative DVOA, meaning on an average play, they do worse than average.

Seahawks are performing 11.4% better on the average play, but that's not evenly divided. We do much, much better on our average offensive play but worse than average on our average defensive play. (Though Seahawks actually had the best defensive DVOA in the NFL over the last two weeks.)

Thanks, this sheds some light. I still do find it curious though. So the 2-5 Jaguars are 5.6% better than an average team per play, but the one loss Giants and Vikings are below that? Interesting

 

[bigskydoc]

... but the one loss Giants and Vikings are below that? Interesting

That means that the Giants and Vikings did worse against their opponents than an average team would be expected to

 

[Welshers]

That means that an average team would be expected to have better results against those teams than the Giants and Vikings did

well it's certainly an interesting metric. it's funny with how abysmal Tampa's offense has been this year, seeing them in the top 10 on average play performance. They've had games borderline unwatchable.

 

[Own The West]

Thanks, this sheds some light. I still do find it curious though. So the 2-5 Jaguars are 5.6% better than an average team per play, but the one loss Giants and Vikings are below that? Interesting

Giants are paper tigers. I’m looking forward to undressing them this week. The Jets too, but that’s a irrational dislike for Robert Saleh thing more that I think the Jets are a bad team.

 

[Welshers]

Giants are paper tigers. I’m looking forward to undressing them this week. The Jets too, but that’s a irrational dislike for Robert Saleh thing more that I think the Jets are a bad team.

Yes I can see this, the Giants have barely been winning. I'm not even necessarily criticizing DVOA as I don't fully understand it, but the metric isn't too convincing to me. I mean, really? The Jags over the Vikings?

 

[Own The West]

Yes I can see this, the Giants have barely been winning. I'm not even necessarily criticizing DVOA as I don't fully understand it, but the metric isn't too convincing to me. I mean, really? The Jags over the Vikings?

Eh, it’s a stat. The only stat in football that tells the whole story is W-L.

 

[HawkStrong]

Giants are paper tigers. I’m looking forward to undressing them this week. The Jets too, but that’s a irrational dislike for Robert Saleh thing more that I think the Jets are a bad team.

?

I actually have a soft spot for the guy, considering he was a part of our SB winning team.