The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball (30 page)

7
. Net defined local revenue basically equals local revenue minus stadium-related expenses.

8
. For a more detailed discussion of these tax rates and redistribution systems, see Zimbalist,
May the Best Team Win
.

9
. The age and characteristics of a team’s stadium will also affect its elasticity. Other things being equal, new stadiums generate more revenue. Other factors might also be relevant, such as the team’s performance the previous year, the existence of star players on the team, and the perception of the team and its owners in the community.

10
. The use of team fixed effects provides for some control over differential market sizes (and branding) across teams, yielding a purer measure of the revenue response to wins at different win percentages. Similarly, the use of year fixed effects allows for the control of idiosyncratic factors in different years, such as a work stoppage or a steroid scandal. The use of the ninth degree polynomial allows for smoothing and a continuous function.

11
. In his book
Diamond Dollars
(Hingham, Mass.: Maple Street Press, 2007), Vince Gennaro purports to have econometrically estimated each team’s win curve, yet he provides next to no information on how he executed his empirical analysis. For instance, Gennaro apparently has done some regression analysis where revenue is the dependent variable and team wins in the current and previous years (which he apparently weighted equally) are two of the independent variables. We are not informed what the other independent variables are, nor are we told whether the relationship is tested linearly on nonlinearly, nor for what years it is tested. The issue that stands out is that there are not enough years of observation to make a statistically meaningful estimate for each team. As encountered elsewhere in the sports analytics world, Gennaro’s invokes the proprietary nature of his work, presumably endowing it with some commercial value. The problem is that his methodology is a black box and cannot be vetted by scientific standards. In his case, the little he does reveal about his methodology inspires little confidence.

12
. Under special circumstances, such as a new stadium, the team’s fixed revenue can be modified. Such modifications, however, have had only a trivial effect on relative marginal tax rates.

13
. To be more precise, each club’s threshold is based on the slots of their draft picks in the first ten rounds of the amateur or Rule 4 draft, wherein each slot has a recommended signing bonus. The sum of these signing bonuses then becomes the team threshold. Since the slots are allocated in reverse order of finish in the previous year’s standings, this generally means that team thresholds are in reverse order of finish. This outcome can be modified when teams lose or gain picks as a result of signing or losing free agents.

1
. DER is also sometimes referred to as defensive efficiency record.

2
. Both marks put Ortiz in the seventy-first percentile among the 248 major league players with at least 1,000 plate appearances between 2000 and 2002.

3
. As we shall see below, from 1998 through 2002 when Dan Duquette was the Red Sox GM, the team had some of the highest saber-intensity scores over the past two decades.

4
. Pedroia was described by scouts as “not physically gifted,” as having a diminutive stature and loopy swing. Writing for ESPN, Keith Law, as late as 2007, opined: “Dustin Pedroia doesn’t have the strength or bat speed to hit major-league pitching consistently,
and he has no power.” Law went on to project the future MVP as a backup infielder. Nate Silver,
The Signal and the Noise
(New York: Penguin Press, 2012), ch. 3.

5
. Silver,
The Signal and the Noise
, also relates that although his player forecasting system, PECOTA, ranked Pedroia as a top prospect, he himself gave up on Pedroia and traded him from his fantasy baseball team.

6
. For example, in eXtrapolated Runs, a single is approximately 47 percent more valuable than a walk; more specifically, a single is worth .5 runs and an unintentional walk .34 runs. The relative value of a single to a walk differs slightly, but not importantly, in other linear weights models.

7
. Oakland’s
onbase
rating in 2002 was 1.034, which placed it in the ninety-third percentile among teams in the past twenty-seven seasons.

8
. For example, computing linear weights via multiple regression on all team-seasons from 1985–2011 with at least 140 games played reveals a coefficient for sacrifice bunts that is negative (–0.0254), but not statistically significantly different from zero.

9
. It is interesting to note that when we run the log of win percentage on the log of our six saber-savvy metrics, the coefficient on (the inverse of)
sacbunt
is negative (and only significant at the .10 level.) This may appear to go against the sabermetric wisdom, but it is important to keep in mind that it is only a measure of successful sacrifice bunts. If we had a measure of all the unsuccessful sacrifice bunt attempts, yielding either an unproductive out or an extra strike on the batter, then the coefficient would undoubtedly be positive.

10
. Since the publication of
Moneyball
, Oakland, Seattle, and Tampa Bay appear to have put a greater emphasis on Defensive Efficiency relative to Fielding Percentage. Seattle’s fascination with defense under Jack Zduriencik is well documented, as is Tampa Bay’s remarkable transformation from the worst-defensive team in baseball to the best. The five leading teams in terms of
der
are as follows: Seattle (2009) 1.031; Boston (2007) 1.030; Tampa Bay (2011) 1.029; Tampa Bay (2008) 1.028; Oakland (2005) 1.027.

11
. Nor is it likely that the saber-savvy GM is trying to minimize FIP relative to ERA. He is just trying to minimize FIP. To the extent that FIP is correlated with more traditional pitching measures, such as ERA, then our index will pick up some part of traditional pitching attributes and overestimate the contribution from saber-intensity.

12
. As we noted in Chapter 3, sometimes ISO is defined simply as SLG - BA, in which case triples are given a higher weight than doubles.

13
. More precisely, the 22 percent of players with between two and three years of major league service with the most major league service accumulated are also eligible for salary arbitration, provided that they had at least 86 days of major league service the previous year.

14
. In our estimates of return to skill for individual players, the player’s labor market status (reserve, arbitration eligible, or free agent) is controlled for with dummy variables for arbitration eligibility and free agency.

15
. As we discuss baseball labor market inefficiencies, it is important to keep a distinction in mind. While walks or on-base percentage may have been undervalued relative to batting average from the standpoint of physical output (runs produced), it is possible that fans find base hits sufficiently more exciting to watch than walks that they are willing to pay more to see a high average hitter than a high on-base percentage hitter. If owners pick up on fan preferences and their revenue implications, then the undervaluation of OBP may be less than it seems from only assessing the impact on runs scored.

16
. The regression model is explained in greater detail in the Appendix.

17
. The increased value of the walk rate during the pre-
Moneyball
period (1998–2002), when power was at its apex, might seem curious. A plausible explanation is that since so many home runs were being hit during this time period, batters who managed to reach base were more likely to be driven in. Thus, the difference between being on first or second was small, since you would be driven in by a home run in either case. Conversely, if home runs were infrequent, then the value of being on second as opposed to first would be much higher. This notion coincides with the “take your walks and wait for the three-run homer” strategy attributed to Earl Weaver, but also espoused by Sandy Alderson.

18
. In particular, the coefficients on OBP, ISO, DER, and FIP are all significant at the .001 level, while baserunning and sacbunts are significant at the .10 level. The equation was estimated with data from 1985 through 2011. In order to use the coefficients as weights, we take their absolute value and normalize them. The econometric details of this and other models are presented in the Appendix.

19
. We normalize all the coefficients and we take the absolute value of FIP.

20
. Stan Kasten, the team’s president, while not an active practitioner, was an avid reader of Bill James and was well aware of the new metrics.

21
. It is interesting to note that when team effects are added to this equation, the only team with a positive and statistically significant coefficient is the Oakland A’s.

22
. In this regression, SI is significant at the .001 level.

23
. We are really estimating marginal physical product (in games) rather than WAR here. Since a replacement player produces 0 WAR by definition, the estimated values are the same.

24
. Silver,
The Signal and the Noise
, p. 107, quotes Billy Beane: “The people who are coming into the game, the creativity, the intelligence—it’s unparalleled right now. In ten years if I applied for this job I wouldn’t even get an interview.”

25
. Bill Shanks’s
Scout’s Honor: The Bravest Way to Build a Winning Team
(New York: Sterling & Ross, 2005) is not a serious discussion of the ingredients of the team’s success. It argues that the scouts were central and that the team favored high school draftees, but it provided little detail or evidence on either account.

26
. Silver,
The Signal and the Noise
, p. 99.

27
. Silver,
The Signal and the Noise
, p. 97.

28
. Silver,
The Signal and the Noise
, p. 99. The minimum salary in 2013 is actually $490,000.

1
. The information used here was obtained free of charge from and is copyrighted by Retrosheet. Interested parties may contact Retrosheet at
www.retrosheet.org
.

2
. Note that this definition of PAY is equivalent to team payroll divided by the MLB average payroll in that season.

3
. We do not show the team fixed effects. It is interesting to note that the Oakland A’s were the only team with a significant (at the .05 level) and positive coefficient. Baltimore, Detroit, the Chicago Cubs, and Kansas City all had negative and significant coefficients.

4
. See
http://www.seanlahman.com/baseball-archive/statistics/
.

5
. Although sabermetricians have corrected for ballpark effects for quite some time, there is little consensus on the most appropriate way to do so. For a fuller discussion see Carl Morris et al., “Improving Major League Baseball Park Factor Estimates,”
Journal of Quantitative Analysis in Sports
4, no. 2(April 2008).

6
. Of course, this is guaranteed by the least squares fitting procedure used in the payroll model.

7
. J. K. Hakes and R. D. Sauer, “An Economic Evaluation of the Moneyball Hypothesis,”
Journal of Economics Perspectives
20, no. 3 (2006), pp. 173–185. Also see, J. K. Hakes and R. D. Sauer, “The Moneyball Anomaly and Payroll Efficiency: A Further Investigation,”
International Journal of Sport Finance
2, no. 4 (November 2007), pp. 177–189.

The index that appeared in the print version of this title was intentionally removed from the eBook. Please use the search function on your eReading device for terms of interest. For your reference, the terms that appear in the print index are listed below

ABPRmetrics bulletin board

adjusted plus/minus (APM)

Alamar, Benjamin

Albert, Jim

Alderson, Sandy

Amaro, Ruben

Anderson, Brady

Antonetti, Chris

Arizona Diamondbacks: effect of park factor adjustments on

employees working in analytics

last-place finishes of

saber-intensity of

Atlanta Braves: effect of park factor adjustments on

employees working in analytics

last-place finishes of

onbase
scores of

saber-intensity of

BA.
See
Batting Average (BA)

BABIP.
See
Batting Average on Balls in Play (BABIP)

Baker, John

ballpark factors

Baltimore Orioles

effect of park factor adjustments on

employees working in analytics

last-place finishes of

onbase
scores of

saber-intensity of

Baltimore Ravens

Barzilai, Aaron

Baseball Abstract
(James)