There are not many times that I would report on a math conference here. For that matter, there aren't too many times when anybody would report on any math conference on any sports blog. However, this is no ordinary sports blog and this is no ordinary conference. I didn't have any strict agenda for this conference, which meant that I could pretty much do whatever I wanted. This meant attending talks on such things as pop-up books, checkers, and knitting, among other things (don't worry, I also attended talks about "real" math while I was there, too)
Of course I was thrilled to find out that today, as well as tomorrow morning, there was a special session on sports math. Other than judging the undergraduate poster session (which was the only real thing I had to do) and going to one other talk, I pretty much spent my entire afternoon there. It started out with a very good talk entitled "Are Umpires Racist?" Since none of you were there, I'll spoil it for you. The short answer is "not in general, but there might be a few who are."
There was one talk, however, that really stood out above the rest. It was given by Ben Baumer who previously worked for the Mets, but has since left to become a professor at Smith College. The talk was about openWAR--an open source statistical package for baseball. He started by discussing what WAR is, as well as what a replacement level player is. WAR is somewhat ambiguous and replacement level is pretty much undefined. He suggested that for replacement level players, we should take out the 750 players in baseball who have played the most, and then aggregate the rest.
There were many other interesting things he discussed, but I won't go into too much detail. You can see the abstract here, and the slides for a slightly different version of the talk here.
After the talk, there was a group surrounding him and he seemed to be talking to one other person in particular, so I nudged my way into the circle. I had to leave before it was over, but later found out the guy he was talking to was the creator of baseball-reference.com! Aren't math conferences fun?
Cool!
I started as a math major in college and wanted to find a way to work in baseball statistics. I was told that was unrealistic so I changed my major. Oh well.
To think that you could have helped billy beane write his book!
It could have been worse. My wife had to choose between healthcare and pension actuary. She chose pension because Hillary was going to takeover healthcare and pensions would be around forever.
Ha ha ha.
Pension obligations will still be around forever.
I lucked into group health actuary because that was the job I was offered, but I'm glad because it fits my interests and talents more than pensions or individual life or investments.
Obama got elected and a big healthcare law that's close to what he wanted passed and I'm busier than ever but guessing a whole lot more. Spending lots of time deciding which dart to throw at which board and which blindfold to put on beforehand.
That part I did help her get right. She took a job with a small local consulting firm 7 years ago. I told her that the companies would look for the lower cost actuaries to maintain their plans going forward.
Her old national firm with offices in the downtowns of every major city have now laid off over half the people she worked with in Minneapolis.
I probably interviewed with them.
I believe they bought me lunch at Palomino.
I ordered paella, the other fellas had nice cuts of meat.
That was the point at which I thought it unlikely I would get the job.
(The only place I didn't get lunch instead hired me, and due to my job, I was able to have some paella on Thursday night about a block away from where Palomino used to be so that was on my mind.)
An open WAR would be huge for the longterm credibility of the statistcal concept. About time.
To be pedantic, WAR is precisely defined: offense + defense + position + replacement. Perfectly open and perfectly useless. I think Baumer knows this, but the linked-to slides seem to conflate the definition with the implementations. Practically speaking, not a big deal, but it is important to be consistent.
Slide 7, "What's Wrong with WAR?", is great. I disagree with his assertion that replacement level has ad hoc definitions. There has been plenty of discussion about defining it at a point that makes sense historically. He might be arguing against the size of the current acceptable range though, so I'm not going to complain too much.
Reading the rest, this is great. I do wish fWAR and rWAR had confidence intervals.
Yes, almost every statistician will tell you that confidence intervals are much more important than mean.
Ooh . . . I want to hear more about the knitting math!
The talk was by Sarah Marie Belcastro from Sarah Lawrence University. Strictly speaking, the talk was about embedding graphs on different surfaces, but she spent a lot of time showing us different surfaces she had knitted (including a Klein Bottle). There was, however, an entire section entitled AMS Special Session on Mathematics and Mathematics Education in Fiber Arts.