An Interview with a Sabermetric Guru Turned Oxford Scientist

Last week I wrote a post and called attention to a great article in Wired about the sabermetric research that Oxford scientists had been conducting to compare players of all eras. In fact, I began the posting with “Finally, Oxford scientists are pulling their weight.” Turns out, that one of those Oxford scientists, Dr. Mason Porter, stumbled across my posting and called attention to it on his blog, Quantum Chaotic Thoughts. I suggest you check it out – it’s likely the only page on the World Wide Web where you’ll find a posting titled “Dodgers acquire reliever George Sherrill” followed by another one titled “Electromagnetic integral equations requiring small numbers of Krylov-subspace iterations.”

In any event, I e-mailed Dr. Porter, who was quoted several times in the Wired piece, and asked if he would take a few moments to answer some questions via e-mail for a post on 27pitches. Dr. Porter came through, and I’d like to take the opportunity to thank him again for his time. Without further adieu, here’s Dr. Porter’s answers to my questions.

1) The marriage between baseball and mathematics has existed since the beginning of the game. However, while the common fan’s knowledge is limited to how many RBI’s Albert Pujols has or what Roy Halladay’s ERA is, it seems that you and your colleagues delve much deeper into the world of sabermetrics. When did this love affair begin, and whose ideas did you fall in love with first – Galileo’s or Doubleday’s?

I definitely fell in love with baseball first.  I grew up in LA and the Dodgers captured my heart early on.  I have some memories of watching the 1981 team, though I was only 5 years old then, so I can’t quantify too much of this.  I remember being saddened with Steve Garvey left to go to the Padres and obviously watching Fernando Valenzuela was magical.  My predilection towards math and science came reasonably early, but baseball was first.  I learned how to do some simple calculations by looking at the statistics on the back of baseball cards, and I sent some pretty obnoxious letters to Topps about both calculation and tabulation errors on their cards.  (Once or twice, I even got a reply.)

2) What’s your beef with Juan Pierre? Was your personal hatred for him a reason to withhold the incorporation of stolen bases into your analysis? Also, what made Bert Blyleven so effective to be ranked ahead of Carlton, Niekro, and Sutton?

Juan Pierre and I have an interesting history.  I haven’t been dissing him as much this year because he’s been playing surprisingly well, and he was actually pretty decent early in his career.  The problem is that he was crappy by the time the Dodgers signed him to a large contract that was ill-founded from day one.  I think that created some bit of resentment on my part, and he’s been one of my favorite targets ever since because for a while he was taking a spot in the starting lineup from more deserving younger players.  I dissed him so much that I even started using a “Juan Pierre” tag line on my blog.  The lack of stolen bases has nothing to do with him.  It was just that our method focuses on pitcher-batter matchups and we wanted to use something simple so that we could convey the effects of the structure of the baseball network (which is the goal of the paper; the rankings are just there to help frame that discussion and are not intended to be taken seriously).  I can’t really give a “why” as to Blyleven was ranked ahead without looking things up.  To answer that, I’d start by going to the statistics and then seeing if who they played might tweak things a bit.  I’d be honored, though, if anything I do helps get Blyleven in the Hall.  It’s really a shame that he isn’t there already—it means that some people are being a bit myopic.

3) Who is your favorite player of all time; favorite player currently (if different), and why?

Nolan Ryan is one of my favorites, and I loved several of the 80s Dodgers—Garvey to some extent, Fernando, Orel Hershiser, (especially) Pedro Guerrero.  [I say 80s even for Garvey just because I only saw him towards the end of his Dodger career.  I know the descriptor isn't fully accurate.]  Other favorites include Greg Maddux because his pitching was so elegant and Craig Biggio because he always got so dirty.  I also have a place in my heart for players like Jim Abbott who really had to overcome a lot.  I’m not sure if I have a favorite player right now.  Maybe Tim Lincecum comes the closest?  I’m not sure.  Some players are just inherently more interesting than others for various reasons, and I gravitate more towards them rather than necessarily which ones are the “best.”  When I was young, most of my favorite players were automatically Dodgers, but that just isn’t true anymore.  The Dodgers are still my team, though, and I would like to believe that they always will be.

4) Do you and your colleagues have any plans to create any new statistics or methods of evaluating statistics in the future?

We do have some plans for that.  The baseball data is rather rich, so it would be nice to do more with that.  We’ve discussed various possibilities, though I think it will be a while before any of them are ready.  We have previously looked at college football rankings, though I’m a baseball fan and that older project was purely an academic exercise for me.  There’s a student working with me this fall who wants to look at cricket player evaluations, so maybe I can get some of the British sports fans mad at me as well.  (Most of my Oxford colleagues don’t realize how serious a matter it is when discussing Pedro Martinez versus Sandy Koufax.) We’ve also looked at rankings in other contexts; for example, we have some projects in which we use the Netflix data.

5) The article in Wired mentions that, in your paper, PEDs were not factored in – here comes the loaded question: is this a fair analysis given this caveat? Do you feel that, because each baseball era is flawed (ie, no blacks in the game early on, alcohol abuse and greenies in the 70s and 80s, PEDs in the 90s and 2000s) that “it’s all relative”? If you had found a way to factor in PEDs (and I’m not sure how you would), do you think that Barry Bonds would still be the top hitter on the list?

We wanted to use the performance as measured by the actual outcomes, regardless of what “extracurricular” activities might have been contributing factors.  As you suggest, there isn’t a way to factor PEDs in without doing something arbitrary.  Ranking systems aren’t robust—small tweaks can have very discernible effects in the final ordering—so one tries to have some justification for the choices one makes.  I wouldn’t know how much to factor PEDs in without taking a number out of a hat.  If I take a sufficiently large number, then Bonds might move down the list.  If I take a smaller number, then he might not.  The point, though, is that we can’t do this without being arbitrary and we purposely wanted to do a study that actively tries to avoid doing such things.  In that event, the best thing is to just say that that is not incorporated in what we studied, while acknowledging that perhaps some matchup results might have been different if it weren’t for them.

6) Finally, what was the most interesting thing you learned by performing

The thing I find most interesting is actually a rather technical point that I don’t fully understand, and it has nothing to do with the rankings. Namely, there is a certain structural feature of the single-season baseball networks that seems to change gradually (and somewhat ‘monotonically’, though I am abusing the term slightly) as a function of time rather than having jumps that coincide with rule or schedule changes.  This involves a relationship between player “degree” (number of players faced, independent of multiplicity) versus player “strength” (number of players faced, including multiplicity) that is in the form of what is called a “power law.”  The exponent in this power law is what changes monotonically in time, which I think is really neat.  I guess it goes to show that even when the topic is baseball, it might be some seemingly-esoteric aspect of the investigation that really fascinates me.

If you’d like to get in touch with Dr. Porter, please visit his blog.