Roundup 10/3

Neil Paine, the new administrator at Pro-Football-Reference.com ranks all QBs in NFL history by their six best seasons. Ken Anderson, John Unitas, Roger Staubauch, Steve Young, and Peyton Manning come out on top.

Neil also points out that Peyton Manning is dominating the Adjusted Yards per Attempt stat so far this year and since 1998. Kurt Warner, Tom Brady, Drew Brees, and Rich Gannon trail. Some surprises (to me) on the list of top 20 QBs include Chad Pennington, Jeff Garcia, and Doug Flutie.

For those interested in the historical comparisons between the AFL and NFL, Jason Lisk puts a bow on his series on draft classes.


Doug Farrar from Football Outsiders has a gig with the Washington Post's 'Smarter Stats' blog. Two of his recent posts are interesting. In one, he tackles the myth of tackles--having a lot of them is not necessarily a good stat for a player. In another, he points out that some teams who aren't doing as well as expected this year aren't stumbling for the reasons most people think.

Football Outsiders also asks if there is such thing as a "clutch" QB. Unsurprisingly, the answer is no. But I actually am a little surprised we weren't treated to a false 'yes.' If you do any kind of real statistical test and use the standard 5% rule for significance, you will typically find about 1 out of 20 cases are significant just by random luck. I see that kind of analysis too often.

Also, the FO clutch analysis uses "late and close" as a relatively primitive definition of "clutch." In my WP graphs, you'll see a stat called "LI" on each play description. LI stands for Leverage Index, which is a quantitative measure of how critical the play is. LI is simply the difference in WP between whether the current play (or series) is a success or failure. The concept is similar to the one used for baseball developed by Tom Tango and others. Right now, I'm not happy with my LI formula and it needs more work before I start using it for anything. The key is to have an appropriate definition of success and failure. (Does failure mean a punt, turnover, sack, etc? Some combination? Does success mean a first down, a TD...?)

Chris from nfl-forecast.com has started posting playoff probabilities already. He bases his calculations on a Monte Carlo model that uses my team rankings. Congratulations to Indy and New Orleans who look like they have already made the playoffs! One bad injury to either Manning or Brees could change those probabilities pretty quickly. It's still very early in the season.

And don't forget about the Advanced NFL Stats Community site. It's a place where anyone can publish his own analysis. I'm happy to post just about anything--stats research, opinion, even (or especially) items critical of my own analysis. If you're trying to start up your own site, cross-posting here might be a good way to get some clicks and some momentum. There's also raw data available to download.

This week at the community site, Luis DeLoureiro has strong feelings about how bad the NFL's QB Passer Rating is. Visit his new football stat site, nflstatanalysis.net

Luis also pointed me this this interesting article from Malcolm Gladwell. In it, he compares how difficult it is to predict NFL QB success with teacher success in the classroom.

Finally, an administrative note: If you have your own site and link to Advanced NFL Stats, please make sure you no longer use the "bbnflstats.com" link. That was the original address that really just started as a placeholder until I could think of a better name. I relinquished the registration to bbnflstats, and sure enough some creep quickly registered it, pirated older material from my site, and put tacky adds all over the place. I was able to convince him to remove my material, but it's been replaced with general football stuff from Wikipedia. Thanks.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

2 Responses to “Roundup 10/3”

  1. Dave M says:

    The article on tackles assumes that they're recorded accurately, which of course they are not. It's an interesting piece nonetheless.

  2. Anonymous says:

    In regards of Leverage Index, what do you currently use for success and failures?

Leave a Reply

Note: Only a member of this blog may post a comment.