The Baseball Economist: The Real Game Exposed (25 page)

The other way in which pitchers can prevent runs is to prevent hits on balls in play. Figure 15 shows that there is almost no correlation between a pitcher’s ability to prevent hits on balls in play, with an R
²
of .06, indicating pitchers have little skill in this area. If fluctuations in ERA are heavily influenced by BABIP, which pitchers appear to have little control over, then ERA contains quite a bit of misleading information regarding a pitcher’s skill.

Because strikeouts, walks, and home runs seemed to be much more stable from year to year than hits on balls in play, McCracken attempted to use these three metrics on their own to predict pitcher ERAs in the following season. In fact, he found that a pitcher’s preceding season’s DIPS are better predictors than that pitcher’s overall ERA from the preceding season. The noise of earned runs generated on balls put in play, which were randomly turned into hits or outs by fielders, actually hindered the identification of the pitcher’s true ability. In turns out that the real reason Greg Maddux is so good is that, though he is not an overpowering strikeout pitcher, he rarely walks batters or gives up home runs. This makes DIPS a valuable tool for disentangling responsibility for preventing runs.

Strikeout rates may be strongly correlated from season to season, but how do well these stats predict pitcher ERAs?

In order to find the answer, we need to establish how accurately DIPS predict pitcher ERAs in the season in which pitchers are playing. Using multiple regression analysis, I estimated the impact of different pitching statistics on ERA. The metrics that most concerned me were strikeouts, walks, homers, hit batters, and batting average on balls in play. At the same time, I included measures of team defense, the age of the pitcher, the league of the pitcher, and the season, to control for these possible outside influences in the predictive model. Table 27 reports the unit and percentage impacts of the variables on a pitcher’s ERA.
⁷¹

The unit impact quantifies the effect of a one-unit change in the pitching statistic on ERA. For example, an increase of one strikeout per nine innings lowers a pitcher’s ERA by about 0.17. The percentage impact (or elasticity) tells us the percentage change of ERA in response to a 1 percent change in the statistic calculated at the average. So a 1 percent increase in the strikeout rate lowers a pitcher’s ERA by approximately 0.24 percent. The percentage impact helps us judge the impact of the different metrics relative to their normal values. For example, the unit impact of every walk (0.30) is nearly twice that of a strikeout (0.17); however, in terms of the average number of walks and strikeouts,
their percentage impacts on ERA are nearly identical (0.23 percent and 0.24 percent).

Differences in the variables included in the regression explained 77 percent of the variance of pitcher ERAs, which is quite good. All of the estimated impacts were statistically significant, meaning there is less than a 5 percent chance that the true impact of a statistic has no effect on ERA. The large impact of BABIP is quite interesting. A one-standard-deviation increase in BABIP (0.023) raised a pitcher’s ERA by about 0.42 (approximately 10 percent of the average ERA for the sample). If pitchers have little effect over balls in play, then a random fluctuation of BABIP can influence a pitcher’s ERA quite a bit. These estimates provide a good baseline to evaluate how well the DIPS predict future run prevention.

Predicting an ERA based on a pitcher’s statistics from the prior season is not much different from the exercise above. Table 28 shows the estimated impacts of these characteristics in the preceding season on the following season’s ERA, controlling for the defense, age, and league of the pitcher. To account for the impact of the defense playing behind the pitcher, I included the team’s batting average on balls in play for
all
pitchers on the team.

Again, all of the estimates are statistically significant, except for hit
batters.
⁷²
The result for strikeouts is quite interesting. The previous year’s strikeout rate impacts the following season’s ERA (0.18) about as much as the season’s strikeout rate affects that same season’s ERA (0.17)—see Table 27 for comparison. This is not totally surprising, because strikeouts are strongly correlated from year to year. While walks and homers are important, they are not as consistent predictors of ERA as strikeouts.

BABIP has a real impact on ERA, but the effect is small and in the opposite direction than one would expect: a higher BABIP is associated with a lower ERA in the following year. Rather than there being some inverse relationship between BABIP from year to year, this is more likely derived from a few extremely high and low BABIP seasons that typically regress to the mean the following year.

But how does the DIPS prediction stand up to a prediction based on the previous season’s ERA? It fares much better, just as McCracken found. Table 29 shows the very weak relationship between the previous year’s ERA and the ERA in the following year. In reality, the effect is nothing, since the estimate is not statistically different from zero. The R
²
is not quite half of the DIPS-only model. Knowing a pitcher’s defense independent stats, sometimes referred to as “peripherals,” does tell us
more about a pitcher’s future ERA than his current ERA. To judge a pitcher solely by his ERA might give him credit or blame for something he’s not able to control. But I’m not sure if these findings necessarily mean pitchers do not have control over balls in play.

Is it possible that pitchers do have the ability to affect hits on balls in play, but that this influence is so strongly correlated with the DIPS that it is masked? Multiple regression analysis identifies a correlation between the predicting and predicted variables included in the model, but it does not tell us why. If a pitcher strikes out a lot of batters, it does not mean necessarily that the corresponding effect on ERA comes solely through the direct impact of strikeouts. The correlation between strikeouts and ERA could reflect a pitcher’s ability to affect hits on balls in play in addition to the direct effect on limiting balls in play. If strikeout pitchers cause weak ground-outs and walk-prone pitchers serve up more line drives, these factors will be captured in the weights assigned to strikeouts and walks in a multiple regression estimation.

For all practical purposes, this possibility is irrelevant—if DIPS tells us all we need to know about run prevention, it doesn’t matter why— but let’s see if it is true. Maybe the previous year’s BABIP doesn’t give us much information about the following year’s BABIP, but the DIPS do, because they are correlated with a pitcher’s ability to affect hits on balls in play. If this is the case, then it’s possible for a pitcher to control BABIP through his DIPS. And in a regression on ERA, the effect would
be captured by the DIPS variables. Table 30 reports that estimated impact of the previous year’s pitching statistics on the following season’s BABIP.

The results confirm something startling in the magnitude and statistical significance of the predicting variables: differences in pitcher control over hits on balls in play are somewhat predictable from past performance. But that information is not in the statistic we would think to look at first, BABIP. This ability has been hidden due to its correlation with DIPS metrics. It turns out that, in fact, the strikeout rates are inversely related to BABIP in the following season. Though it’s not widely discussed, Voros McCracken also found correlations between both strikeouts with a pitcher’s future BABIP. The effect for strikeouts seems a bit obvious. The fear of strikeouts possibly induces hitters to take weaker protective swings to stay alive, and thus yields softer hits that are more likely to result in outs.

But just because something is statistically significant does not mean it is practically significant. Using the estimate of the impact of the predicting variables reported in Table 30 and the earlier estimate of the impact of BABIP on ERA (18.16) in Table 29, I am able to assign an earned run value to strikeout prevention through its effect on balls in play. For every one strikeout increase per game, the BABIP decreases by 0.00172. Multiply that by 18.16 and every strikeout saves 0.03 earned runs per game through reducing hits on balls in play, which is not meaningless but small. Why doesn’t BABIP correlate very well from year to year when strikeouts do? Well, there’s just a lot more noise generated by random bounces from year to year in BABIP than there is in strikeouts; therefore, it’s hard to directly observe pitcher control over this metric.

Although it turns out that pitchers do seem to have some minor ability to prevent hits on balls in play, it does not alter the predictive element of DIPS theory one bit. Why not? Because that ability is captured in DIPS statistics.

DIPS theory is a powerful tool in evaluating pitchers independent of their fielders. More importantly, the story of DIPS is a triumph of intellect over an intractable problem. Michael Lewis quotes McCracken, responding to the claim that separating the pitching and fielding contributions in run prevention was impossible: “That’s a stupid attitude. Can’t you do
something
? It didn’t make any sense to me that the way to approach the problem was to give up.”
⁷³
The joint prevention of runs would be quite complicated to sort out if not for this imaginative invention. Alfred Marshall—the father of modern neoclassical economics, wrote

The economist needs the three great intellectual faculties, perception, imagination, and reason: and most of all he needs imagination, to put him on the track of those causes of visible events which are remote or lie below the surface.
⁷⁴

He would applaud the inspired work of Voros McCracken.

13

What Is a Ballplayer Worth?

—LEO DUROCHER, FORMER MANAGER