Levitt pretty consistently embarassed himself on the Freakonomics blog about a year ago, if I remember correctly, regarding this subject. While he's a real smart guy, there's also a lot of real smart baseball stats nerds out there who were refuting him. It got to the point where it seemed like he was arguing back "just to argue." His operating thesis was, in short, that the A's had similar OBP's to the Red Sox and Yankees during the time before Moneyball was written. What he was missing, unfortunately, was that the first part of the title is "money-" because the batters performed similarly despite being paid far less than the Yankees or Red Sox lineups.
The discussion and comments might still be archived on the Freakonomics site; I deleted it from my bookmarks around the time of the posts regarding Moneyball because I figured, if the author could blindly stick to his misinformed assumptions on something like this - where I am knowledgeable enough to know he's wrong - I am afraid of all the other instances in which he's misinforming a reader like me and I don't know enough to call him on it.
Hmm. We operate in a funny "science" don't we? With all due respect to Skip and Jahn (and my respect is considerable), theirs is but a first step in the direction of determining the validity of Lewis' conjecture. For example, Leavitt's alternative wasn't tested directly by Sauer/Hakes; maybe pitching matters too. Of course, then the question becomes why the A's would have such good pitching with such a low payroll.
Dave is right that it is interesting why Beane's strategy worked; the existence of such innovation is always interesting. But they are interesting because we can test whether they were a failure or just a transition with new "market making" or if the value of the strategy only just beats transactions costs, or some such.
But at present we neither know that "it took baseball more than a century to learn the lesson of plate discipline" nor that "it took the publication of a best-seller to correct the market imperfection."
Indeed, we don't even know that it was market imperfection until alternative explanations are all examined.
For me, I'll watch for future tests that add to the Sauer/Hakes findings about OBP.
Back in the late 1980s and early 1990s, the old internet newsgroup rec.sport.baseball was full of people talking about the importance of using OBP and SLG (instead of the triple crown stats) to assess the marginal physical product of batters in baseball. It was tonnes of fun.
Many of those same folks have progressed to interesting careers in sports and sports management; one example can be seen in the work "baseball prospectus" that directly emerged from the work people were doing back then and talking about on the internet; others from that same era are doing background work for many major league teams. At the time, MANY of us wondered why on earth some smart GM didn't hire a few of these folks and use the information they were producing. Billy Beane clearly found some discovery value. He (and Skip's work) showed that the efficient markets hypothesis is correct in the long-run, but the adjustment sure isn't instantaneous.
Meanwhile, several posters back then noted that Branch Rickey at an early stage in his career thought that using (Hits + BB)/(AB + BB) was a pretty good measure of batting effectiveness.
Given that Rickey knew this, and given that we rec.sport.baseball regulars knew that OBP was important, Dave's question becomes even more interesting: why did it take so long for the importance of OBP to even approach getting into mainstream usage? I honestly don't think it has made it yet; too many announcers and baseball folks still say "moneyball" with a bit of a sneer in their voices.
As one of those early proponents of using OBP and SLG, when I did play-by-play for the AA London Tigers in the early 90s, I refused to mention RBIs, and I consistently and constantly explained to the various co-announcers that I worked with why OBP and SLG were so important. Then in the mid-90s, when I did play-by-play for the independent league London Werewolves, I actually had it written into my [puny] contract that the screen stats for each player had to include OBP and SLG and not RBIs.
Some of the people I worked with have gone on to work with MLB teams. They certainly understood the importance OBP before Moneyball was published. I think they even regretted its publication, just a little bit, because it gave away some of their secrets.
The focus on OBA and OPS is, I suppose, not surprising. But Lewis's contention was that Beane had found a skill that was undervalued, and he used that to build winning teams. The specific skill that was undervalued happened to be OBA (in Lewis's and Beane's opinions).
"Moneyball" is not pursuit of high OBAs. It's pursuit of undervalues skills.
So here's a question. If Hake and Sauer are correct that OBA is no longer an undevalued skill, are there other underevalued skills out there for intelligent management to exploit?
I have a (bad?) habit of trying to tie loose threads together when I get a chance. So far,
1. hypothesis test of A's pitching + where did it come from? (Rod Fort, reminding us of Levitt's hypothesis)
2. Why didn't this catch on much earlier, given Branch Rickey, Bill James, BP-guys, etc.? (ElectEcon and JC)
3. Are there other undervalued skills (a.k.a. whither now?) (Donald Coffin)
The following is just my conjecture, and it is possible Skip will disagree, but my (untested) opinions on these items are:
1. The A's took advantage of the fact that there are three distinct markets for players: the spot market for free agents, the monopsonistic market for (drafted) minor leaguers, and the market to trade players between franchises. An improvement on the Sauer/Hakes JEP paper would separate the free agent and draftee markets fully (rather than using the d.v. fixed effect) and study returns to skill dimensions for players acquired as free agents relative to those developed in-system. [Note: I need to think more about players acquired by trade. If still under reserve clause, there are still rents available, but how the rents would be split between the trading franchises is not obvious to me.] The update would similarly evaluate dimensions of pitching skill.
For "expensive" inputs, the A's acquired players with the draft and developed them internally, so the high market prices didn't affect them. If anything, the lower internal price was a source of rent for them. A testable (but to my knowledge, still untested) hypothesis is whether the A's were statistically superior at drafting and developing prospects during the entire Alderson-Beane era. They DID have several highly productive draftees on their playoff teams (Giambi, Tejada, Chavez, Zito, Mulder, Hudson, etc.). To the extent that rents provided by these drafted players made team success through adjustments on the OBP margin possible, Alderson may not have been given sufficient praise in Lewis's book.
Importantly, the A's relative appreciation for OBP doesn't mean they DISLIKE slugging. They were, and are, more than happy to draft players with power potential if they can somehow sign the player for a smallish bonus.
The Moneyball hypothesis, as Don noted, is more than just about OBP (that was just the example in this case study). More generally, when the A's went on the spot market for players, they looked either for guys with undervalued skills, or guys in the "irregulars" bin: players with blots on their record that reduced their market value, but with enough upside potential in the GM's estimation to be worth taking the risk. (Note: You can still see signs of this philosophy in the '06 A's when looking at Frank Thomas [good eye; poor recent track record] and Milton Bradley [reputation of being... difficult])
If the A's were drafting pitchers based on a sabermetric model that predicted developmental success better than the conventional wisdom does, t
[shoot. 3000-character limit kicked in.]
... If the A's were drafting pitchers based on a sabermetric model that predicted developmental success better than the conventional wisdom does, that would be a separate source of rent. For example, perhaps they felt that college pitchers with high K/BB ratios and low OPS against were more promising prospects than high school pitchers with blazing fastballs and high K/IP ratios. An analysis of their draft picks with Davenport-like translations on college/prep stats might turn up such proclivities. [Q: has anyone done this?]
A misunderstanding that concerns me is that some readers seem to think that there is a need to pick sides. It is entirely likely that BOTH Levitt and Sauer/Hakes could have valid points. Skip and I focused upon the aspect of buying low and selling high as improving the team greatly at relatively little expense; Levitt prefers to focus on the creation of rents through the draft (albeit realized in the form of wins rather than money).
These are not mutually exclusive hypotheses. We've tested ours. I'm sure someone will test Levitt's sometime soon (if someone hasn't already). My guess is that the A's indeed were drafting "smarter" than other franchises, but that still won't invalidate the JEP paper. If I get a chance after class, I'll post again to put down my guesses as to what the "next" round of undervalued skills might be (to answer Don's question).
An aside about the last paragraph of the comment by JKH: who was it who said, "Good pitching beats good hitting...... and vice versa"?
Donald Coffin asks the (cogent) question "What skills are undervalued today?"
I think that the baseball skills that are undervalued are defense and speed. Again, look at the A's - their recent teams have not featured the high-OBA slugs at every position that the 2000-2003 A's did - the A's have been among the league leaders in Defensive Efficiency the last few years.
More generally, most sabermetric types would agree that evaluation of individual player's defensive ability is one of baseball's Hilbert Questions (Keith Woolner of Baseball Prospectus wrote such an article). We can assess whether a team is good at turning balls-in-play into outs, but it still seems pretty murky to decide whether Derek Jeter or Michael Young is a better defensive player.
059ba0 fcebd4eb7f
Commenting by HaloScan
