Major League Pitching Performance Is Poorly Predicted By Prevailing Metrics

Abstract

As scientific analysis continues to supplant subjectivity in sport drafts, the predictors of pitching success in professional baseball remain improvable. PURPOSE: To evaluate the pitching performance of first-round draft picks in Minor League (MiLB) and Major League (MLB) Baseball. METHODS: 1,000 MLB pitchers drafted between 1965 and 2014 were selected for analysis; all players had a minimum of 3 complete MLB seasons. There were 362 starters, 315 relievers, and 80 closers; 243 served multiple roles. We evaluated MiLB and MLB performances of all pitchers and tested the effect of draft pick on outcomes using linear regression models, holding potential confounders constant. Dependent variables were: earned run average (ERA), ERA+, runs allowed per 9 innings (RA9), winning percentage (W%), fielding independent pitching (FIP), and strikeouts per 9 innings (K9). Analyses were conducted on the total sample as well as players drafted after the year 2000 (N=396). RESULTS: There were 404 pitchers drafted in the first round; draft round number ranged from 1 to 69 (mean: 7.3 ± 9.8). Mean years in MiLB were 10.1 ± 4.4 and mean years in MLB were 8.0 ± 2.8. Across the total sample, holding constant draft year and pitcher class (e.g., starter, reliever), first-round picks had 0.233 higher ERA in MiLB (p<0.001; 95% CI: 0.153 to 0.313), 0.386 higher ERA in MLB (p<0.001; 95% CI: 0.304 to 0.469), a 0.217-point increase in MiLB RA9 (p<0.001; 95% CI: 0.129 to 0.306), a 0.398-point increase in MLB RA9 (p<0.001; 95% CI: 0.308 to 0.489), a reduction in MiLB W% by 2.239 percentage points (p<0.001; 95% CI: -3.393 to -1.085), and a reduction of 2.435 percentage points in MLB W% (p<0.001; 95% CI: -3.466 to -1.405). Additional MLB metrics predicted by first-round draft status were a reduction in ERA+ by 8.304 points (p<0.001; 95% CI: -10.333 to -6.275), a 0.244-point increase in FIP (p<0.001; 95% CI: 0.175 to 0.312), and 0.302 fewer K9 (p=0.002; 95% CI: -0.492 to -0.113). In the subsample analysis, controlling for pitcher class, the direction of every relationship was preserved. Significance was retained in MiLB ERA (p=0.016), MLB ERA (p<0.001), MiLB RA9 (p=0.078), MLB RA9 (p<0.001), MLB ERA+ (p=0.002), MLB FIP (p=0.001), and MLB K9 (p=0.045). CONCLUSIONS: The qualities that predict a first-round draft pick for MLB pitchers do not predict success in MiLB or MLB.