Writers of Pro Football Prospectus 2008

01 Jan 2013

ESPN: Does Momentum Mean Anything?

Do teams that are hot entering the postseason tend to outperform teams that played well all season long? Evidence seems to suggest that it doesn't matter.

Posted by: Vincent Verhei on 01 Jan 2013

14 comments, Last at 03 Mar 2013, 1:55am by Perücken


by HumanitarianDave :: Tue, 01/01/2013 - 1:49pm

Bill Barnwell posted about this same topic on grantland and came to a similar conclusion.


by jklps :: Tue, 01/01/2013 - 2:08pm

Barnwell worked for FO before getting hired by Grantland, and references DVOA in many of his articles.

by Hummingbird Cyborg :: Tue, 01/01/2013 - 2:24pm

I don't doubt this is true, but I wonder about overlap. Many teams with a DVOA advantage also have a similar weighted DVOA advantage.

The sample size of teams who are significantly better in weighted DVOA is likely tiny, so no real conclusion can be made. Too bad though. It is curious, but even if there was a statistical conclusion, it would be bound to be weak and we all know examples in which a tangible physical cause produced the change.

But this article does enough to say that it's silly to have an idea that peaking matters because if that were true, it would be easy to see.

by Special J :: Tue, 01/01/2013 - 4:26pm

I was wondering about the same things, as well as the possible distorting effect of home-field advantage. DVOA will likely correlate more strongly with a better seed than weighted DVOA. If the playoff win% were adjusted for HFA, would you start seeing more of an effect?

Of course, I think the biggest problem is sample size, especially once you start focusing on situations where there's a larger disparity between regular and weighted DVOA. How does the predictive value of DVOA and weighted DVOA compare in-season?

by Bright Blue Shorts :: Tue, 01/01/2013 - 2:54pm

Can't read the article as I'm not signed up ... does it examine the opposite view - how teams that don't have momentum do?

by essay (not verified) :: Wed, 01/02/2013 - 7:34am

Thanks for this cool article dude!

by joebarnin :: Wed, 01/02/2013 - 1:45pm

I'm not a insider so I can't read the article, but I did a study on this a couple of years ago, and came to the conclusion that momentum going into the playoffs is a myth: http://home.comcast.net/~jeffreyeby06/momentum/Momentum.htm. This was first published in the Pro Football Researchers Association newsletter.

by JIPanick :: Wed, 01/02/2013 - 4:53pm

Frankly, I always thought y'all should scrap weighted DVOA and just use the full version for all purposes.

I mean, your sample is only 16 to begin with; thinking you can get better results by trimming it is kind of silly.

by LionInAZ :: Wed, 01/02/2013 - 6:29pm

Weighted DVOA doesn't trim the sample. It just gives a higher weight to recent games than those at the beginning of the season. Comparing weighted DVOA vs unweighted DVOA tells you if a team has been performing better or worse than early in the season. That's all.

by JIPanick :: Thu, 01/03/2013 - 3:20pm

I was under the impression that it ignored games outside the ten most recent in addition to the weighting. In any event, I still think that DVOA and not Weighted DVOA should be used in the playoff odds report and game picks.

by nat :: Fri, 01/04/2013 - 11:39am

I actually have no clue what Weighted DVOA means any more.

But from this 2004(!) DVOA article, I saw this:

The importance of games 10-12 weeks ago has been reduced, the importance of games 13-14 weeks ago is really small, and games 15 weeks ago and earlier now don't count at all.

And the link from the glossary goes to an even older article, that has a note saying Weighted DVOA was changed in 2004. I assume the change was the one I found. If it's changed since then, I hope Aaron retrofits a link into that old article, or links the glossary entry to a newer one.

Like you, I had vague memories of a ten game cut-off. Like you, I prefer to use DVOA and adjust for injuries etc on my own. But that's a personal preference.

by LionInAZ :: Fri, 01/04/2013 - 10:22pm

I agree, it would be helpful if they could clarify the definition of weighted DVOA.

by bstar :: Wed, 01/02/2013 - 6:14pm

Where's the link to the actual article?

by Perücken (not verified) :: Sun, 03/03/2013 - 1:55am

Frankly, I always thought y'all should scrap weighted DVOA and just use the full version for all purposes.
I mean, your sample is only 16 to begin with; thinking you can get better results by trimming it is kind of silly.