The 2011 Football Outsiders Almanac was released last week and, as is the case every year, it’s a must-read for fans of the Green Bay Packers and the NFL. Mike Tanier wrote the chapter previewing the Packers upcoming season and focused a good portion of his preview on GM Ted Thompson.
Adam Czech: Do you think Ted Thompson knows what DVOA is? If not, how do you think he would react after you explained it to him?
Mike Tanier: Thompson probably does not know what DVOA is. Every time we have spoken at length to a coach or GM about our methods, we have gotten reactions like this: “We do something similar, though the calculations are different, and it is based off our game film and internal methods.” The coach and GM know exactly what play was called and what each player’s assignment is, so they can analyze film in ways no one outside the organization can. One thing we always hear is that we are doing the right thing by making our stats situation-dependent: coaches want different things on 1st-and-10, 3rd-and-15, and 3rd-and-inches in the red zone, and DVOA reflects those differences.
AC: We’ve seen some major changes in the methods MLB and NBA GMs use to evaluate players and build teams over the last 10 years. Will Thompson’s team-building strategy — finding value players off the scrap heap that fit his coach’s system and provide depth — become the newest NFL team-building trend? Or have teams always been using the Thompson method, but we just haven’t noticed because a) Nobody is as good at the Thompson method as Thompson is or b) Few organizations have the patience to actually see if the Thompson method works?
MT: There is a team that does what Thompson does better than Thompson: the Steelers, who have followed a similar model for 20 years. The best organizations have done what Thompson does for many years. The Eagles do it to a degree, as do the Patriots, and once you get past Peyton Manning and the top five or six guys on the Colts roster there are lots of street free agents and castoffs.
The Steelers do something slightly different than the Packers, in that they are more likely to horde 6th-round picks on the bench for two years and develop them, while the Packers have been plucking guys off of other team’s practice squad. That practice squad market, the “reserve futures” market, is a bigger deal for some teams than others. One GM told me that he always preferred getting another team’s practice squad player to getting a rookie free agent or another 7th round pick: this guy, after all, went through some training camp and survived some cuts. I think a few teams will see what Thompson did and perhaps become more aware of reserve-futures. Remember: it is no substitute for drafting a franchise quarterback and a superstar pass rusher, just a supplement.
AC: Can statistical analysis help identify players like Frank Zombo, Sam Shields and Howard Green? Or will finding these types of depth players that fit a system always be more about good ‘ol fashion scouting and film study?
MT: At Football Outsiders, we have a system called SACKseer that can be used to identify collegiate pass rushers. The system is not perfect, and no statistical scouting tool ever will be, but it is designed to organize a defender’s college sack totals, experience, and Combine numbers into one number you can make sense out of. Identifying safeties, cornerbacks, and interior linemen will always be the realm of scouting informed by stats, not stats themselves. At some point, scouting and stats merge: if I want to create a stat for defensive tackles called “double teams fought off,” and I watch every game the tackle played in and count every double team that didn’t blow him back five yards, well, I am really scouting, aren’t I?
AC: Clay Matthews is the only returning player on the Packers front seven that had more than 6.5 QB hurries last season. Should the Packers be worried?
MT: To a degree. Jenkins was a loss. Remember that Matthews is a special, special player, and that because of injuries and rotating players, a lot of effective players did not produce huge numbers of sacks or hurries. The Packers are a team that can get seven hurries or sacks/hits from seven different players over the course of a game.
AC: The Packers have successfully used a no-huddle offense during each of their first three preseason games. Which is more important if a team wants to run the no-huddle: A talented and cohesive offensive line or a stud quarterback?
MT: The quarterback must be comfortable in the no-huddle. The no-huddle must be something that got the full attention of coaches and players in meetings, walkthroughs, and camp. The exact quality of the personnel matters about as much as it matters in any other element of the offense. When you see a team running a sloppy no-huddle, often it’s because it was implemented as something “off-to-the-side” in their offense.
AC: The Packers actual wins have been fewer than their Pythagorean Projection three years in a row. Why?
MT: The Packers had severe penalty problems in 2008 and 2009 as I recall, and they lost a bizarre number of games by close scores, usually at the end of the game. They often had problems icing wins: they didn’t run the ball very well, and opponents often got two or three possessions late in the fourth quarter. That being said, luck can play a huge part, because three years is just 48 games, which is really a small sample size. Flip a coin 48 times and you will probably not get 24 heads and 24 tails. We always research to see if we are over or under-valuing some specific part of football performance, but there are some things that are almost impossible to responsibly input into a statistical method, like most penalties, and other non-predictive stuff.
A big thanks to Mike Tanier for taking the time to answer these questions and provide some additional insight on the Packers. In addition to the Football Outsiders, Tanier writes for the New York Times Fifth Down Blog and is an excellent follow on Twitter.
Adam Czech is a a freelance sports reporter living in the Twin Cities and a proud supporter of American corn farmers. When not working, Adam is usually writing about, thinking about or worrying about the Packers. Follow Adam on Twitter. Twitter .
17 thoughts on “Talking Packers With Mike Tanier From Football Outsiders”
Nice interview, Adam. Thanks for posting.
I love that site, and profootballfocus as well. It is a good place to go that gives you a different look at the performance of players. Great for fantasy football as well. But there are times on those sites where the statistics can be kind of misleading, just like in any statistical analysis. I like that those sites are geared basically to fanatics (you know…like us) and there is no fluff on them. Hard to find that information. I wish the NFL network would cater more to the fan that is REALLY into the game and is knowledgable. It seems like they started that way and now are just like espn and all the other crap out there. Lets break down cam newton for the 198th time shall we?
You’re spot on about statistics sometimes being misleading. This is especially true in football because the sample size is so small.
It always bothers me when the “stat geeks” and the “traditionalists” bicker with each other, though. There’s no one right way to evaluate players. There’s plenty of room for statistical analysis and traditioanl boots on the ground scouting. Both sides get along better today than 10 years ago, but there’s still some work to do.
I pretty much agree with this.
What I’ve found, because I’m a stat junkie, is the more limited the focus of the statistic is, the more accurate the statistic itself is.
You can divise a pretty accurate way to tell which QBs are better on 3rd downs. You can formulate a precise way to determinate which pass rushers are better in clear passing situations.
But there’s not a single statistic out there that can precisely say who is the best QB or DE/OLB in the league.
Statistics are great situational tools, but if one tries to reach a broader conclusion based solely on statistics, it’s seldomly accurate.
Nice work Adam. Tanier’s a smart guy and a good person to talk to about football. I need to get my hands on that almanac now!
It’s a must-buy for me every year.
Reminds me of the old joke:
Boss needs the answer to a problem. What’s 2 + 2?
Engineer: 2 + 2=4.0000000000000000000000
Personnel: I need to check the Equal Op Regs.
Statistician: Pulls shade, locks door, checks for bugs and says in a whisper, “What do you want it to be?”
Just saying Stat Modles are only as good as the variables input and no model has ever been successful at prediciting human behaviors. At best the results are directional. Add physical actions and you end up with a hodge podge of predictions that are meaningless.
It’s a lot of fun to read and review their stuff. But, if a coach started using it as part of their evaluation and decision making proccess, I’d start pushing for a new coach.
Any kind of evaluational tool that doesn’t take into account a player’s assignment is ultimately flawed.
Disagree big time. If a coach didn’t use all tools available to him when evaluating players, I’d want a new coach. Stats and analytics are part of that tool box and should be used in conjunction with traditional scouting, film study, gut feeling, observations, assignments, etc., etc.
My disagreement is w/ Ron 🙂
If that’s so there would never be a failed new product would there? The variables used in these models are highly subjective and subject to bias. Hence, the output is questionalbe.
I don’t get your argument. Because something isn’t 100 percent accurate, it shouldn’t be used?
Statistical analysis is a tool that I am sure MM and TT already use. Most coaches especially offensive coaches are detail nuts. I am sure they are looking tons of statistics in their game planning and all facets of team management. Pliars are not always the tool needed for the job, but that does not mean it isn’t a good tool.
Identifying patterns – Yes. Predicting future behaviors – No.
Sorry Adam. just got a burr up my butt over know-it-all statisticians. Especially those who trying to hard sell their product. He’d have better luck trying to sell “sham wow.”
No need to apologize. But I don’t get where your dislkike of statistical analysis comes from. Using stats to identify patterns helps you predict future behaviors. It doesn’t predict future behaviors — nothing does — it just helps you. It’s another tool to try and get ahead.
Comments are closed.