Aaron Schatz, the founder of FootballOutsiders.com, will be the lead author of Football Prospectus 2005.
Aaron Schatz: Hi everyone. Gee, weren't we here just three days ago? Anyway, I once again would like to thank Baseball Prospectus for allowing us to use their chat setup to talk some sabermetrics, football style (or as Pat Laverty calls it, "safermetrics"). Just a reminder -- I'm the Football Prospectus guy, so save your baseball questions unless they are questions about how baseball stat analysis and football stat analysis are similar/different. The BP guys are super busy writing the book so they're letting me sit in with the band for a session. I'm back to full health, I've got Kaveret and Jorge Ben ready in my stereo in honor of questions from Israel and Brazil, and lunch by my side, so let's rock.
Chris Hartjes (Toronto): Hey Aaron, do you check out other fantasy websites to see what they are up to and try to incorporate some of the things they do into your work?
Aaron Schatz: Well, first of all I should say that I don't consider Football Outsiders to be a "fantasy website." Like BP, our goal is to analyze the actual game on the field -- the values of players, the comparative quality of teams, strategies for improving your franchise and going to the Super Bowl. It has ramifications on fantasy football, sure, but that's not the major thrust. Like BP created BP Fantasy, we may also create some more fantasy-focused content next season.
As far as reading fantasy websites, I do like to hunt around various message boards around the Web to see what folks are talking about. There is one fantasy football writer who in my opinion is way ahead of everyone else when it comes to using objective analysis in sabermetric style to write commentary on fantasy football. His name is Doug Drinen and he writes for footballguys.com. You might recognize him as co-author of the recent paper which theorized that since pitchers don't hit in the American League, they do not fear retaliation like their counterparts in the NL. He also runs pro-football-reference.com.
Gordon (Albany OR): The peak ages for baseball hitters is thought to be about 27. What about for football players? And by position?
Aaron Schatz: And speaking of Mr. Drinen, he happens to be the person who has done research on this question. Because Doug's focus is fantasy football, his research was solely on "skill positions." He discovered that the age at which players begin their decline is generally 26 for RB, 28 for WR, and 30 for QB. We have not done research on other positions yet, although I am guessing that the peak age is younger for defensive backs and older for punters, kickers, and linemen. But this is a guess only.
What's interesting about the peak age study Doug did is that this year has turned it upside down. Many of the best players this season at RB and WR have been players who are past their peak age. For WR, that means Owens, Muhammad, Horn. For RB, Martin, Dillon, Barber, and for crying out loud what the heck rejuvenated Jerome Bettis? The question becomes: is this a fluke? Or will peak age move later in the coming years? In many sports we are now finding players sticking around and even performing at a very high level far longer than what we previously expected. Barry Bonds, anyone? Or, for those who question the, um, legality of his home runs, Roger Clemens and Luis Gonzalez?
Israel (ISRAEL): Do your newfangled statistics take into account knees at the ends of games? For instance, the Steelers have had seven or eight games where they could have scored more points but took knees at the end - including both the eagles and the patriots with first downs inside the ten. Does this credit them with anything? (Or does it show up as a failed red-zone visit?)
Aaron Schatz: Hey, here is our Israeli friend. In how many other countries would you get a guy whose name is the same as his country of origin? Anyway, I count neither kneels nor spikes in my statistics. This is one of the reason why our numbers will differ slightly from official NFL statistics. I also try to manually go through the play-by-play and mark Hail Mary interceptions as "HM," which then gets counted similar to a normal incomplete.
e (Chicago): I'm rooting for you in your www.twominutewarning.com matchup, but I would like to ask you about one matchup in particular...
You avoided Chicago-1.5 v Houston, which I thought you'd jump on Houston
Chicago dellusional in calling Hutchinson a true quarterback
would you please be kind enough to share your thoughts?
Aaron Schatz: e is referring to the TwoMinuteWarning.com head-to-head game picking contest run by my buddy Roland Beech. I'm in the semifinals this week. I did avoid picking this game because of the infamous Anthony Brancato system, which demonstrates that dome/warm teams have trouble in cold weather late in the season. If forced to choose every game this week, I would favor Houston, but since I only had to take five for the contest, I skipped this one.
I should note that if you want to read Anthony's article, go to FootballOutsiders.com and search for "cold weather betting." Also, I forgot to say in the previous answer that Doug's article on peak years is available in the Articles section of pro-football-reference.com.
dantroy (davis, ca): While Eli is young and is certainly suffering from a lack of support, doesn't his historically awful performance bode poorly for his future projection? What do you see as his future?
Aaron Schatz: Hey, are you reading my mind? This is going to be the subject of the "intro" section for next Monday's Snap Judgment at ESPN.com. But I'll give you a sneak preview since you have to suffer with living in Davis. (I kid! I kid because I love.)
I have a little Bill James-esque similarity scores system, it is no PECOTA but it has its uses. I pro-rated Eli's first few games and ran the similarities. The good news is that the #1 most similar rookie eventually got a Super Bowl ring. The bad news is he had to go to another team to do it. His name is Doug Williams. Vinny Testaverde also shows up in the top ten. And, um, a lot of garbage. But it is tiny sample size, and my similarity scores have absolutely no adjustment for strength of schedule, so I'd say that hope is certainly not yet lost. Read more on Monday.
lk6 (Santos, Brazil): Hey Aaron.
I'm 16-yr old Steelerfan from Brazil.
My question is: Do you think that the Eagles can beat the top AFC teams? (PIT, NE, IND, SD)
Thanks in advance.
(Sorry for my poor english :( )
Aaron Schatz: Alrighty, Kaveret album over, Jorge Ben album goes in, so let's answer our Brazilian question. Umbabarauma! Homen Gol! As you may know, lk6, I live in Framingham, MA, and half the population of this town is Brazilian.
The Eagles have to be heavily favored in the NFC playoffs, but they will face a team in the Super Bowl far better than anyone they've played this year except the Steelers. Can the Eagles win a matchup with those teams? Sure. But if those games were played today, I would favor the AFC team in each one except perhaps San Diego. Michael David Smith this week wrote about the Eagles' run defense in his Every Play Counts column. It is not as bad as last year -- Jeremiah Trotter has been a major part of the improvement -- but it is still the team's weakness, and Hollis Thomas being injured is not going to help things. Each of those four AFC teams you mentioned has a very strong running game. I also think that Pittsburgh and New England (assuming Law is healthy by then, which he should be) also should have defensive backs that can keep Terrell Owens from going completely nuts. The Colts might actually be the best matchup for the Eagles, because they have the worst defense of the four top AFC teams. If that's the Super Bowl, take the over.
Joel (Washington, DC): The playoff match-up I really want to see is the Steelers defense, with its incredible strength and quickness, against the awesome Colts offense. I understand from a previous chat that you rank the Colts right now as the best team in football, but do you have any "scientific" way to predict the outcome of such a confrontation? If the Colts do meet the Steelers, and assume that Plaxico Burress is healthy and able to play, what's your take right now as to the likely result?
Aaron Schatz: Actually, I don't rank the Colts as the best team in football right now. Total DVOA, which ranks teams by their play-by-play for the entire season, has New England on top. Weighted DVOA, which gives more strength to recent performance, has Pittsburgh on top. Indy is #3 in total DVOA and #2 in weighted DVOA. To be honest, the top four teams are all so bunched up and ahead of the other 28 that they have to basically be considered even.
As far as a "scientific" way to predict the outcome of such a confrontation, I have no all-in-one measure that puts a ton of variables into an equation and spits out a winner. One of the important things to remember about football, as opposed to baseball, is that every playoff round is Game 7 and Game 7 only. People want definites -- "Philadelphia WILL beat Atlanta" but all you can really give are trends, tendencies, probabilities. There are too many random events that can happen in the course of one game to guarantee anything.
So what I can do is what is in your final question, tell you the trends and tendencies. Newsflash: Manning is out of his gourd. On the other hand, Pittsburgh's pass defense has truly come from nowhere to be the best in the league this season according to DVOA. And what is important for a matchup with Indy is my little method for splitting defensive DVOA against various types of receivers. Pittsburgh is #7 against number one receivers, #2 against number two receivers, and #4 against number three receivers, which is an indication that they should have less of a problem than most teams with the Colts' biggest strength, that if you cover one weapon Manning just finds another. Of course, the Steelers like to run and run some more, and the Colts have a weak run defense.
And most importantly, the importance of that first week bye in the playoffs cannot be overstated. Nor can home-field advantage, especially when you have a dome team visiting a cold weather team in a game that has a 75% chance of being played at 4pm or later. That is the main reason why the Colts have only a slim chance of beating either Pittsburgh or New England, even if we consider them even in quality.
mscogle1 (Seattle, WA): Aaron: The Seahawks are stumbling towards the playoffs but face a challenging offseason, with key offensive players Hasselbeck, Alexander, and Walter Jones all free agents. Of the three, who would you attach the franchise tag to, and if you could only keep two of the three, who would you say bye to?
Aaron Schatz: I actually wrote a piece about this for the Sun a couple weeks ago but we ended up editing it into a piece more about the current Seahawks, with the idea that I would wait until the offseason to comment on free agency.
Left tackles may be the most important players in the league. Jones has actually been a free agent in each of the past three offseasons as well, and each time the Seahawks have placed the "franchise player" tag on him, and every year it annoys Jones enough to make "Walter Jones holdout" an AutoText item in the word processors of all Seattle beat writers. I don't see that changing next year.
Hasselbeck I really don't know about. One of the problems with being in the infancy of football analysis -- I mean, we're 25 years behind the baseball guys -- is the difficulty in separating one player's performance from his teammates. Who is Matt Hasselbeck? The guy from last year? The guy from this year? Somewhere in between? Quarterbacks are hard to find, just ask Arizona or Cleveland. I think a lot of it depends on what Hasselbeck is asking for.
That leaves Alexander. As many people know, I am an advocate of the running back by committee for a number of reasons. One reason is that RBBC often is just as effective as most of the single star running backs in the league. A second reason is economic: a RBBC usually costs less than a single star running back, and for less money you get injury insurance to go with similar production.
I wrote a longer piece about this for Slate a few months ago: http://slate.msn.com/id/2106074
(At the end I say that Miami will go as far as their passing will take them. And I was right. Their passing took them nowhere, Ricky Williams or no Ricky Williams.)
So I would franchise Jones, sign Hasselbeck as long as he doesn't want to be paid as a top five quarterback, and let Alexander walk. Check our stats at FO and you'll notice that Maurice Morris and Alexander have pretty much the same DVOA.
The other issue here comes one year later, when they'll be stuck having to pick between Jones and Steve Hutchinson, their other great lineman, and will only be able to franchise one.
BTW, mscogle had another question I can answer quickly...
mscogle1 (Seattle, WA): Has there been much work done on evaluating points scored/points allowed ratios to predict W/L records in football, similar to the "Pythagorean" approach used in baseball?
Aaron Schatz: Yes. The exponent is 2.37 instead of 2. The work was done by a guy named Daryl Morey. Teams will deviate from their Pythagorean records to a greater extent in football, simply because of the small sample size, but it is a good guide to things. The team with the top Pythagorean record has won the Super Bowl something like 16 of the last 20 years (don't quote me on that, I might not be remembering correctly, but the actual number is somewhere on Football Outsiders). This was actually one of the first articles written at FO when the audience was like four of my fraternity brothers and the guy in the next cubicle at Lycos:
One interesting note this season is that PHI, NE, and IND all have better Pythagorean records than PIT. (I ran that a week ago, and I don't think Week 14 changed it.)
ManU Red Devils (Toronto): i love your work on footballoutsiders. good luck with your bp association. i was interested in doing some analysis on the value of field position in different down situations. Is there a public source with compiled historic game logs, or will i have to go though them individually?
Aaron Schatz: This was also a question in the latest FO mailbag: http://www.footballoutsiders.com/ramblings.php?p=2100&cat=3
That link tells you how to find pbp from 2001-2004. Everything before that is off the Web, alas. I'm trying to hunt down as many years as possible.
If you have a specific research question in mind, email me at aaron-at-footballoutsiders.com with the details and when I have time perhaps I can get you the cut of data you need and you can write the research for us as a guest column. We love guest columns!
And we are super psyched to be affiliated with BP, if I have not mentioned that 15 times already.
Patton1941 (Charlottesville VA): Is there a metric that measures the performance of college players and either translates those stats to the NFL or could be used as a predictive performance tool at the NFL level?
Aaron Schatz: I've used the words "holy grail of football statistics" to describe three things.
1) Being able to separate individual players almost entirely from team performance, including rating the quality of defensive players and offensive linemen.
2) Being able to create a Doug Pappas (z"l) style metric that measures how much value a team is getting from its salary cap distribution, and doing real economic analysis of how much players should be paid based on position and performance.
3) The above.
I guess there can't be three holy grails, and since I'm Jewish I should really be calling these something else (The Torah, Prophets, and Writings of NFL analysis? Whatever.) but someday, man, someday. Part of the problem is that finding consistent NCAA pbp logs is impossible, and you would also need to figure out how to adjust for the fact that every team is running a different offensive style in a conference that plays a different general style, and so forth.
Unfortunately, you won't be getting sweet translations that tell you what prospects would do in the NFL in Pro Football Prospectus 2005 like you get in BP. Tune in for PFP 2013 maybe.
Joe E (Trenton): Do you think football analysis will progress to the point where a vertical passing game with virtually no running game will be shown to be statistically superior to traditional balanced attacks?
Aaron Schatz: Nope, because it isn't. I've done some research here that I never had a chance to write up but it probably will be in the book. (Plug! Plug!) Balanced teams generally fare better than unbalanced, except that balance is not 50-50, it is something like 55-45 in favor of passing more.
TGisriel (Baltimore): Looking forward to the Colts-Ravens game, leads me to wonder whether your site has data on the effect on the game of playing on an artificial surface like the Colts. Also how do you treat the newer generation artificial turf like the Ravens'?
Aaron Schatz: That's on the to do list. You should see the to do list. It's massive.
DavidCrowe (Canada): My father bleeds Honolulu Blue and I'd love for him into get into your methodology - my only fear is what you would say about the future of Joey & his beloved Lions.
Aaron Schatz: This is the question that I promised I would answer from Tuesday's aborted chat. As I noted, Michael David Smith is the Lions expert. We were talking about this the other night a bit, I'll try to remember everything he said. The offensive line is improving since the beginning of the year. Kevin Jones has a lot of promise. Roy Williams and Charles Rogers have to figure out how to stay healthy.
Joey Harrington is the big question, isn't he? You don't want to give up on a guy too early after what the Chargers are going through this year. The tough thing about Harrington is that he has been just good enough to keep hope alive but not good enough to show he's definitely a qualified starter. At least with someone like Cade McNown, Akili Smith, we knew right off the bat those guys were not going to work. But for three years now Harrington has stumbled along at a level just below mediocrity. Similar to Tim Couch, I guess, and at some point the Lions have to decide what to do. Mike thinks they need to bring in a veteran who will seriously challenge him next year to either fix his bad habits or lose his job. The worst is that at the first sign of pressure he throws the ball away. I mean, sometimes he throws it away without a sign of pressure at all. Mike McMahon is the opposite of course, constantly trying to make plays that aren't there.
I'm not sure what to say about the defense. DVOA says they are about average against the run and the pass. Mike's been impressed by Big Daddy Wilkinson, who he thought would be a failure. I'd recommend asking him further questions at mike-at-footballoutsiders.com.
As far as getting into my methodology, just like the guys at BP one of my hardest tasks is explaining what I do and how it works in the simplest possible terms to make things accessible. I am always looking for people to send us constructive criticism of how we can make our statistics easier to understand, so don't hesitate to do so.
OrioleDog (Reisterstown, MD): Did the Colts overpay for Harrison?
Aaron Schatz: I'm still trying to figure that out. I don't remember where I said this but trying to understand NFL contracts is like trying to read Martian. The years and dollars are all completely fake, and the contracts are filled with clauses that change things relative to the salary cap. I'll let people know if I ever figure out what the heck is going on here.
(This is a call to anyone out there who wants to become the FO salary cap expert. Seriously, email me.)
The one thing I'll say is that a) Harrison has declined for the last two years, although his peak was so ridiculously high that his decline still leaves him as a great receiver (baseball equivalent: Pedro if he did not have a torn shoulder) and b) remember that thing I said about guys playing better this year past what should be their peak? Well, Harrison isn't one of them in 2004, he has declined from last year, but that trend might make me less reticent to sign him long term.
lexomatic (Toronto): two questions:
1) relating to Gordon's peak age question. are there any characteristics linking the over the peak players good performances (i.e. health, style of offence, varied skill set -RB's who can catch- etc)?
2) i remember someone asked about the CFL in the last chat. Considering there have been so many succesful QB transplants from the CFL, isn't it worthwhile looking into some kind of translation for the league? have you guys looked into this at all?
Aaron Schatz: Seriously, I'm loving all the Canadian questions.
1) Another one on the to do list, but one thing you might notice about the running backs this season who are playing well past their peak is that none of them have been recently overworked. Dillon had a lot of rest last season, Bettis has been a part-timer for a couple years, Barber runs less often than most feature backs and took a couple years to become a starter, and Martin has Jordan to take some of the load off him.
2) To do list again. Like the first Japanese translations that Clay Davenport did, you would mostly be depending on guys going from here to the CFL, not the other way, and the problem with that is that the guys going from here to the CFL probably had extremely limited playing time. But I watched the Grey Cup and Henry Burris (ex-Bears) looked a lot better than some of the backup QBs who have been forced into starting roles this year.
TGisriel (Baltimore): Speaking of balance - What are your thoughts on balance on a team between offense and defense. Again thinking of the Ravens and the Colts (it's a Baltimore thing)these two teams are perhaps the teams who have shown the least balance over the last few years. Any thoughts on whether the Ravens paradigm (defense rather than offwnse) is better than the Colts paradigm (offense rather than defense?
Obviously being good on both is better, but....
Aaron Schatz: I wrote about this a few weeks ago on FO and got to expose my Philip Glass fetish: http://www.footballoutsiders.com/ramblings.php?p=1872&cat=1
Look at my DVOA ratings for this year and you'll notice that, at least in 2004, the offense-oriented teams (IND, NYJ, GB, MIN) are far superior to the defense-oriented teams (BAL and BUF but also CHI, WAS). That's probably just a one year thing though.
What is clear, however, is that defensive performance varies from year to year more than offensive performance. That would seem to say that building the team on offense is better than building on defense if you have to choose, because you have a better chance of year-in, year-out quality performance. Notice to the Seahawks: I did not say "skill players," I said "offense."
Aaron Schatz: A couple final words to questions I couldn't answer. Joe, I honestly don't do it much. e, email me and I'll try to answer. jrmayne, the 49ers have much more of a clue about their future direction than people think.
Thanks again to everyone for reading, and thank you to the Baseball Prospectus folks for allowing us to use their chat setup. I invite everyone to visit FootballOutsiders.com on a regular basis and to pose questions to both me and the other writers in our discussion threads and through email. We hope you find our analysis as informative and accurate as BP's. I'll probably be back here the week before the playoffs start for another chat.