I’ll note that the Sports Reference blog Statheads now has this blog on their sidebar, something I’m quite grateful for. Not that you don’t have to fight for readership as an amateur in football blogging, because you do. But to be spoken of in the same breath as sites like Football Outsiders or Advanced Football Stats is, well, heady stuff. So to Neil Paine, thank you.
What I’m going to say to readers that are largely team (e.g. Dallas, Atlanta, Green Bay, Pittsburgh, Chicago) fans, you’ll get the best bang for your buck by looking at the tag cloud on the right of the blog and clicking ones that interest you. If you’re one of the football coaches that have drifted here because of Coach Hoover’s recommendations, I’d suggest your best usage of this site would be to follow the tags “46 defense” and “defensive fronts”. For those for whom algebra isn’t an issue, just read the general flow of this board. I’m going to try and keep the look visual and fold most of the numerical results behind “more” tags. The guy who wants to see Rob Ryan or Dom Capers get after the quarterback doesn’t need screen shots of program output, and the guy who does can click on the “more”.
The big splash of the draft, from my point of view, was the trade from 27th to 6th by the Falcons. They netted Julio Jones with the trade, giving up two firsts, one second, and 2 4th round choices in the process. This is because the Falcons felt they weren’t explosive enough, and so had to improve on an offense ranked in the top ten in the league. That their defense was mediocre and the salient feature of the last Super Bowl was that the #1 and #2 ranked defenses met, seemed to be bypassed in the quest for game changing explosiveness. I suspect trying to become a 21st century Air Coryell certainly has fan and box office appeal, but is it wise over the longer term? I’m reminded of the nursery rhyme:
For want of a nail the shoe was lost.
For want of a shoe the horse was lost.
For want of a horse the rider was lost.
For want of a rider the battle was lost.
For want of a battle the kingdom was lost.
And all for the want of a horseshoe nail.
I’m not convinced there was that much more air based explosiveness to get out of the Falcons offense. Perhaps on the ground, where some rest for Michael Turner might save him from regression to the mean.
JMO, but the focus of the Falcons is classic Parcells style ball control, where yards per carry are far less important than time of possession (this, incidentally, is why Curtis Martin will always be underrated by YPC-heads – Parcells just never cared about his ball carriers YPC). In such an offense, the most important component of the offense are first downs, not pretty stats. Given how light the Falcons defensive line tends to be, keeping them off the field as much as possible has to be a serious design consideration for the whole team.
We’ll remind our fellow statheads of this result:
And do some counting, since counting is what statheads do. If you break the teams on the chart into three categories, those below 7.5 picks per year, those between 7.5 and 8.5 picks per year, and those above 8.5 picks per year, how many teams with a winning percentage of 55% or more are there?
|Picks/Year||Total Teams||50% wins or more||55% wins or more||Super Bowl wins|
|7.5 – 8.5||17||9||2||2|
More points to make. There is a modest negative correlation between first round draft choices and long term winning. There is even stronger negative correlation between long term winning and top 10 draft choices. But to note, there really is no correlation between any particular kind of draft choices and winning, and the first words out of Thomas Dimitroff’s mouth to the local radio stations was that he had, at the time, 6 draft choices left.
To amplify on the logistics results of my previous post, I’ll note that if you split the 2001-2010 data set into different kinds of playoff wins, the effect of previous playoff experience disappears – it’s not resolvable if you get rid of too many data points. It had a p = 0.03ish with 110 playoff wins. More careful examination of my data set (from nfl.com) showed 5 home Super Bowl wins in 10 games. I’m not sure how nfl.com is assigning home teams to Super Bowl wins, but that clarification was in order.
I’m tempted to suggest that previous playoff experience goes away with the first week of playoff results, but with so few “week 18″ games, there isn’t enough data for me to be happy claiming that. The strength of schedule result gets stronger the deeper into the playoffs you go. Some useful plots are given below:
Playoff home field advantage: in 100 non-Super Bowl games, there were 60 wins by home teams. This suggests there are really three components to the logistics model of my previous post, not two. And within the context of the model I’m presenting, there really is no measurable effect of regular season victories that isn’t somehow folded into the home team playoff advantage of roughly 60 percent.
Finally, if you’re a newbie to logistic regression, the little pamphlet above comes highly recommended. Parts of it are available via Google Books. Read it and you’ll get a very good grasp of how to interpret these kinds of curve fits.