August 2, 2007
Lies, Damned Lies
Baseball's New Underclass
In an essay for the forthcoming It Ain't Over--which you should all have in your hands in a week or so--I wrote about the permanent underclass of baseball teams that existed more or less continuously from the end of the deadball era to the start of World War II. These teams--the Phillies, Senators, Browns, A's and Boston Braves--combined between 1919 and 1943 to have just 32 winning seasons and two World Championships in 125 tries. My argument is that this wasn't merely the result of incompetent management (though surely that played a large part in their failures). Rather, the economic incentives of the time, particularly the prevailing practices in the scouting and development of young players, tended to favor the rich over the poor and reinforce any differences in the standings.
I was reminded of this when the Red Sox' trade for Eric Gagne came over the wire, for that was another case of the rich getting richer. The trade embodied all of the things that make the Red Sox so dangerous: they have the financial wherewithal to pay the balance of Gagne's contract, the excess prospects to trade, and the shrewd GM to execute the move and steal a player out from under his rival. Those are basically the three inputs in your long-run baseball production function: brainpower in the front office, manpower in the minor leagues and the power of the purse.
Right now, there are five teams that are head and shoulders above their competition in the American League--the Red Sox, Yankees, Tigers, Indians, and Angels--and they all appear to have at least two of those advantages working in their favor. How frustrating it must be for the Royals when the Tigers draft Rick Porcello, or the Orioles when they see the Yankee farm system getting stronger every year. Before, there were two reliable ways for the financially-challenged teams to compete: they could go the old-school route and outscout them, or the Moneyball route and outthink them. But now, the big guys are doing that stuff at least as well as the little guys.
So just what will the American League look like five years from now? Let's return to the three potential sources of comparative advantage that we discussed a moment ago: financial, scouting and management resources.
Among our three categories, only the financial component is immutable. Dayton Moore can scout his butt off, or execute the best trade since Jeff Bagwell/Larry Andersen, but he can't make Fortune 500 companies move to Kansas City, or reverse the emigration to the Sunbelt. Now, certainly, teams can do things to improve their revenues; winning is the most obvious one, and there are the understudied areas of public relations and marketing. Even with that, it's easier to scrape dollars out of your market in New York than Kansas City.
I did a rather involved study of market size earlier this year, the idea of which was to get at the true degree of advantage that each market provides, rather than necessarily the fraction that the present tenant is exploiting. Where I left things a little undone was in exploring the relationship between market value and winning. The Yankees have, by my calculations, a media market that is effectively 80 percent bigger than their next closest competitor in the American League (the Los Angeles Angels). Does this translate more or less linearly into their degree of comparative advantage, or are there diminishing returns involved? In a very large city, for example, you might have more fans who could potentially watch baseball, but there's also more competition for the entertainment dollar. There are also some institutional disincentives in the form of revenue sharing and the luxury tax.
There is a graph of my market-size numbers against the average number of games won between 2000 and 2007. The relationship between the two factors does appear to be approximately linear, although we do ever so slightly better with a logarithmic function that accounts for diminishing returns. Based on the linear version, the Yankees are 3.04 standard deviations above the league average in terms of their market size. With the adjustment for diminishing returns, on the other hand, they are "only" 2.43 standard deviations above the norm. Here are the market size z-scores (standard deviations above or below the mean) for all of the American League teams.
Market Size z-scores with diminishing returns adjustment
Team z-score ------------------ NYA +2.43 LAA +0.90 BOS +0.49 TOR +0.33 CHA +0.23 TEX +0.08 DET +0.05 BAL +0.05 OAK -0.06 SEA -0.35 CLE -0.35 TBA -0.61 MIN -1.37 KCA -1.83
These numbers are intended to be read in terms of the degree of competitive advantage that each market conveys. For example, the difference in market-based resources between the Yankees and the Red Sox is nearly four times larger than that between the Red Sox and league average. It's at either end of the bell curve that these differences matter; the disparity between a "big" market like Boston and a "small" market like Cleveland is mostly perception.
Scouting and Development
One ambiguity in evaluating the scouting and development capabilities of each franchise is that there is an under-accounting for their relative opportunities. Consider this: did the Rangers' scouting and development capabilities improve when they traded for Jarrod Saltalamacchia and Elvis Andrus earlier this week? Not really. Their farm system improved, certainly. But the trade doesn't tell us anything about their inherent capability to mine baseball talent. (If Saltalamacchia and Andrus develop into stars, that might tell us something.)
I mention this because if you look at the team with the strongest farm system in baseball--which is probably the Devil Rays--it turns out that they have had just about the most favorable draft position of any team in the majors since 2000 (I'll explain what I mean by this in a moment). And if you look at the team with the weakest farm system, which is probably the Astros, they have had just about the least favorable draft slots. Since our goal is to evaluate the long-term core competencies of each franchise--we are hoping to look at process rather than results--it is worth trying to account for this distinction.
Rany Jazayerli's outstanding series on the amateur draft contained some data on the relationship between draft position and the long-term expectation of a particular player. What Rany found is that the value of draft picks tails off rather dramatically after the first dozen or so selections, and that the first pick in particular is worth quite a lot. We can use regression analysis to fit a curve to his data. The third pick in the draft is expected to produce about 38.1 WARP in the first 15 years after he is selected, for example, and the 38th pick about 5.4 WARP. What we have, in essence, is a baseball version of the NFL draft value chart (and it turns out that the relative values of different draft slots in baseball and the NFL are quite similar).
I used this chart to project the value of every pick that a team had in the 2000 through 2006 drafts, up through and including the 100th selection. The Royals, for example, had a cumulative "Jazayerli Value" of 320.0, more than twice that of the cross-state Cardinals:
Jazayerli Value of Draft Picks, 2000-2006
Team Value ---------------- KCA 320.0 TBA 311.0 DET 286.6 BAL 270.2 CLE 261.2 MIN 260.6 COL 254.6 SDN 251.5 PIT 251.2 MIL 251.1 WAS 249.3 CIN 235.7 CHN 229.3 FLA 222.6 TEX 214.1 OAK 213.9 TOR 205.5 ARI 201.4 ATL 189.8 NYN 183.5 CHN 182.9 LAA 180.0 LAD 178.5 BOS 168.3 SLN 149.3 PHI 140.1 NYY 138.2 SEA 135.9 SFN 135.1 HOU 119.2
I then compared these numbers to the PECOTA Upside scores of the young players within each organization. Players aged 25 and younger (who almost certainly were drafted no sooner than 2000) were given full credit for their scores, while 26- and 27-year olds were given half-credit. I'm working from the Opening Day version of the PECOTAs, so we are not accounting for anything that's happened since then.
It turns out that the relationship between the two variables is not all that strong, though I don't know how much signal we should expect given all the sources of noise in this data. Many of these players, after all, are still developing, and young players can come into an organization through mechanisms other than the amateur draft.
Nevertheless, the relationship is significant enough that we can make some adjustment for draft slot value. The idea is to compare the actual PECOTA scores to those expected from a team's draft position. The Blue Jays, for example, have an expected Upside score of 1322 but an actual one of 982, so their scouting and development capabilities are evidently rather poor. This plus/minus score can then be standardized into a z-score, as I have done below.
PECOTA Upside Team Expected Actual +/- z-score ----------------------------------------------------- TBA 1595 2503 +908 +1.77 MIN 1465 2159 +694 +1.32 LAA 1256 1910 +654 +1.23 CLE 1466 1916 +450 +0.80 NYA 1148 1535 +387 +0.67 BOS 1226 1348 +122 +0.11 SEA 1142 1109 -33 -0.22 OAK 1344 1269 -75 -0.31 DET 1532 1412 -120 -0.40 TEX 1344 1089 -255 -0.69 TOR 1322 982 -340 -0.87 KCA 1619 1272 -347 -0.88 CHA 1264 815 -449 -1.10 BAL 1490 894 -596 -1.41
Front Office Management
If market sizes and farm systems are difficult to get a handle on, then surely it is a fool's errand to try and quantify the value present in a team's front office. Nevertheless, I did so the best way I know how, which was to appeal to my colleagues. I asked each BP staffer to assign a simple score of 1, 2 or 3 to each American League GM, of "1" meaning above-average, "2" average, and "3" below-average. I received a total of 18 ballots, including my own.
The front offices broke down fairly naturally into different tiers, which in some sense was my goal. I didn't want people to try and pick apart every nuance about whether Andrew Friedman or Terry Ryan is a better GM, but I feel reasonably confident in our ability to see the big picture without falling victim to groupthink. Boston, Cleveland, Detroit, New York and Oakland rated well across the board, while Baltimore, Seattle, Texas and Toronto rated poorly. Everyone else was in the middle. In some cases, like Minnesota, this was the result of combining a lot of 1's and a lot of 3's, while in others like Kansas City, nearly everyone gave the GM a "2". The complete rankings are presented below, along with the z-scores and a comparison to the ESPN SportsNation poll on the subject.
Team GM Rating z-score ESPN -------------------------------------------------------- BOS Epstein 1.13 +1.22 2 DET Dombrowski 1.13 +1.22 6 OAK Beane 1.18 +1.13 1 CLE Shapiro 1.29 +0.96 7 NYA Cashman 1.34 +0.87 3 LAA Stoneman 1.77 +0.17 8 TBA Friedman 1.87 -0.01 13 MIN Ryan 1.88 -0.01 5 KCA Moore 1.92 -0.09 10 CHA Williams 2.02 -0.27 4 TEX Daniels 2.45 -0.97 11 TOR Ricciardi 2.60 -1.23 9 SEA Bavasi 2.70 -1.41 12 BAL Flanagan 2.80 -1.58 14
Now, strictly speaking, it can be argued that a GM is not a long-term asset; he is only as valuable as his current contract. Nevertheless, most front office hires are reflective of the prevailing organizational culture. We have seen a macro paradigm shift toward more analytically-inclined GMs over the past decade, but at the micro level, this process has operated gently; perhaps only in Tampa Bay and Chavez Ravine did the change in culture come abruptly. Moreover, these guys stick around longer than you might think. The average tenure of the current American League GMs is 6.2 seasons, as compared with 3.7 seasons for the current AL field managers.
Triple-Z Score (Long-Term Organizational Power Rating)
Our payoff--the Triple-Z score--is simply the sum of an organization's z-scores in the categories of financial, scouting, and executive resources respectively.
Team Financial Scouting Mgmt Triple-Z --------------------------------------------------------- NYA +2.43 +0.67 +0.87 +3.97 LAA +0.90 +1.23 +0.17 +2.30 BOS +0.49 +0.11 +1.22 +1.82 CLE -0.35 +0.80 +0.96 +1.41 TBA -0.61 +1.77 -0.01 +1.15 DET +0.05 -0.40 +1.22 +0.87 OAK -0.06 -0.31 +1.13 +0.77 MIN -1.37 +1.32 -0.01 -0.06 CHA +0.23 -1.10 -0.27 -1.14 TEX +0.08 -0.69 -0.97 -1.58 TOR +0.33 -0.87 -1.23 -1.77 SEA -0.35 -0.22 -1.41 -1.98 KCA -1.83 -0.88 -0.09 -2.81 BAL +0.05 -1.41 -1.58 -2.94
The American League teams do in fact break down rather definitively into different tiers, although they are not precisely the tiers that we had anticipated. The Yankees form a tier unto themselves. The difference between their score and that of the second-place Angels is larger than that between the Angels and the seventh-place club, the A's; those teams demarcate the boundaries of the second tier. Bypassing the Twins, who are as hard to pin down as ever, we then come to six teams that are in danger of forming the new underclass. With the possible exception of the White Sox, who are fading fast from the elite group but not hopelessly out of reach depending on what Kenny Williams decides to do this winter, it's hard to envision any of these teams competing for a pennant any time soon. They are not only behind in the race, but they are also slower runners.
The term "competitive balance" has been all but vanquished from baseball's vocabulary; it appeared in conjunction with "baseball" just six times in The New York Times archive in 2006, as compared with 47 times in 2002. Nevertheless, while the National League retains its parity, it appears as though the rich/poor gap in the AL is only widening. When this season concludes, the Rangers will have gone eight years without a playoff appearance, the Orioles nine, the Blue Jays 13 and the Royals 22. While a return to the conditions of the 1930s is unlikely, the industry needs to be prepared for the eventuality that there will be a vocal minority of disenfranchised teams the next time the labor agreement comes up in 2011, and some deflating franchise valuations before then.