The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball (24 page)

Our first task is to understand the role of team payroll upon performance. Our data set contains performance statistics obtained from Retrosheet,
¹
as well as team year-end payroll data (in nominal dollars) obtained from MLB’s Labor Relations department and used with permission. The data contains complete information on all major league clubs (768 team-seasons) from 1985 to 2011.

Next, we want to contextualize the payroll data. We do this by defining PAY as the share of a team’s payroll relative to the league average share, which is by definition the total amount spent divided by the number of clubs. Note
that this correction is necessary, since the number of clubs has not remained constant over the time period in question.
²
Thus, PAY implicitly controls for both U.S. inflation and for baseball salary inflation simultaneously. In order to characterize the relationship between payroll and winning percentage, we run a regression model for winning percentage (WPCT) as a function of PAY, PAY
²
, and team fixed effects.
³
The details from this model are shown in
Table 20
.

Table 20. Relationship Between Win Percentage and Payroll, with Team Fixed Effects

Our model satisfies the conditions for multiple linear regression well. Although the quadratic term was not statistically significant at the 5 percent level, we chose to include it so as to incorporate the desirable notion of diminishing returns to payroll. Moreover, analysis of the residuals favored the model that included the quadratic term.

In the previous section we described a model for a team’s winning percentage (WPCT) as a function of its relative share of league payroll (equivalently, the ratio of their payroll to the league average). In this section, we develop a series of metrics designed to measure the intensity of a team’s sabermetric practice. That is, we will attempt to quantify the extent to which the on-field performance of a team reflects sabermetric thinking.

In what follows, we develop relatively simple metrics that we hope will capture some element of sabermetric practice on the part of front offices. Generally, our approach to building metrics that measure sabermetric intensity is to examine the ratio of a saber-savvy metric to a more traditional performance metric. In this respect, we are not advocating for any of these particular metrics, but merely suggesting that they might be popular among
either traditionalists or sabermetricians. In practice, this process involves several steps:

1. Apply ballpark corrections to performance data

2. Compute the ratio of each statistic to the league average in that year

3. Compute the ratio of the sabermetric statistic to the traditional statistic

In the resulting metrics, a higher score reflects more saber-intensity, i.e., better performance in the sabermetric statistics relative to the traditional statistic.

First, we have to adjust our raw performance data for ballpark conditions. We have chosen to do this using the ballpark factors provided by Sean Lahman
⁴
(appropriately scaled). These park factors reflect inflation in
runs
that are attributable to a specific park. For example, the batting and pitching
run
park factors (BPF and PPF, respectively) for Fenway Park in 2004 were 1.06 and 1.05, respectively. This suggests that Fenway Park inflated run scoring by about 5 to 6 percent, relative to a league average park. Simply dividing each statistic by this number would result in statistics that were correlated to the park factor. In contrast, the notion of correcting for ballpark is to
remove
the component of a statistic that is attributable to the park. Thus, we correct for park by building a simple linear model for each statistic as a function of the park factor, and then adding the resulting residuals of that regression to the league average. This procedure removes nearly all correlation between the park factor and the park-corrected statistics.
⁵

After correcting for ballpark, we want to correct for the run scoring of the time period. This has changed considerably over the time period in question, as demonstrated in
Figure 13
. We thus normalize the metric relative to the average value of that statistic for each league in each year.

Figure 13. Runs Scored per Game, 1985–2011

Each dot represents one team in one season, with the league average shown as a line. It is apparent that run scoring was at its highest level during the late 1990s, and that it has fallen sharply in recent years.

To illustrate the nature of these corrections, we compare OBP, park-factored OBP (OBP.pf), and relative park-factored OBP (rel.OBP.pf) by franchise over this time period in
Table 21
below. Note how the park factors affect a few teams in extreme ways (e.g., Colorado, San Diego), while interpreting the statistics relative to the league average improves the standing of the National League teams considerably.

On the pitching side, we see similar changes. While teams like the Dodgers and Mets have low ERAs overall, after correcting for their pitching-friendly ballparks, their standings decrease. Conversely, teams like Texas and Colorado that play in hitter-friendly ballparks move up in the rankings.

Table 21. Effect of Park Factor Adjustments on OBP, 1985–2011

The central idea behind our sabermetric intensity metrics is to examine the ratio of a statistic that likely reflects sabermetric adherence relative to a traditional metric that purports to measure that same quantity. For example, adherents of sabermetrics are more likely to value OBP over batting average
(BA), so proponents of sabermetrics are likely to have higher ratios of OBP/BA than teams that value traditional metrics.

Table 22. Effect of Park Factor Adjustments on ERA and FIP, 1985–2011

To illustrate, the following scatterplot shows the relationship between relative OBP and relative BA, with Oakland’s teams highlighted in black. Thus, in considering a team’s on base ability, we define
onbase
to be this ratio.

Figure 14. Scatterplot of
onbase
, 1985–2011

Each dot represents one team in one season, with the horizontal coordinate given by the team’s park-corrected batting average relative to league, and the vertical coordinate given by the corresponding figure for OBP. The black diagonal represents a ratio of 1:1. Oakland is shown with dark black dots. The fact that these points lie above the diagonal in all but three seasons reflects an emphasis on OBP relative to batting average.

While this is an admittedly crude measurement of sabermetric adherence, it does implicitly control for the quality of players that a team is able to put on the field. That is, while the Royals’ OBP may not keep pace with that of the Yankees, there is a priori no reason to believe that their ratio of OBP/BA should not be as high.

It will likely come as little surprise that Oakland dominates the list of teams that score highly in this metric, claiming five of the top twenty spots over the period 1985 to 2011.

We construct similar metrics that value each of the basic elements of baseball: getting on base, hitting for power (
iso
), pitching (
fip
), fielding (
der
), baserunning (
brun
), and sacrifice bunting frequency (
sacbunt
). For all of these
metrics, the league average in any given season is 1 by definition. We believe that while these sabermetric intensity metrics are far from perfect, they do capture something meaningful.

Table 23. Top Twenty Team-Years in OBP/BA (
onbase
)

We saw previously that Oakland had the largest positive team effect in our payroll model. In this context we want to understand the deviations from the general relationship between winning percentage and payroll, so we compute the model without team fixed effects, and later investigate the relationship of those residuals for patterns with respect to our composite saber-intensity (SI) metric. The equation tested below is WPCT = f(PAY, PAY
²
).