I published a series of FanPosts at the start of the season that graded the 2012 minor league performances of virtually all pitchers in the Astros organization who had thrown mostly in a starting role that season. The system compares the starters to peers while ignoring run allowance, hit and extra-base hit allowance, and other factors that are affected by environment (league, home ballpark, climate, etc.), defense (fielding skill, defensive positioning, etc.), and luck. Instead the system focuses on 4 statistics that are significantly correlated with pitchers’ run allowances but are only marginally influenced by environment, defense, and luck.
The primary motivation is to grade the 2013 performance of the arms who have pitched in the tandem rotations of Lancaster and Quad Cities and faced 150+ batters through games of June 16, given that each squad has reached the virtual halfway point of their regular season with their league observing a 3-day All-Star break. A secondary goal is to quantify how each has improved or worsened relative to 2012.
These are the 4 statistics evaluated:
- BB&HBP %: Walks plus hit batsmen, divided by batters faced.
- K %: Strikeouts divided by batters faced.
- LD&OFFB %: The percentage of nonbunted batted balls that are line drives or outfield flyballs. Slugging on batted balls (and run allowance, logically) increases as this increases.
- OFLD&OFFB Pull %: The percentage of outfield line drives and outfield flyballs that are hit to the batter’s pull-field third of the field. Slugging on batted balls increases even more as this increases.
Performance scores are generated for each stat based on how many standard deviations (SD) better or worse the player has been this season versus the mean of same-handed "starters" who pitched at the same level during 2012 (it’s a bit early to be computing means and SD for 2013, so I’m effectively assuming that the level average and SD won’t change dramatically from 2012). So a lefthanded pitcher in Lancaster’s rotation will be compared to lefthanded starters who faced at least 150 batters in a 2012 High A league (California, Carolina League, or Florida State). A 50 score indicates level-average performance, scores above 50 are better than level-average, and any 10 point change in the score amounts to 1 SD. An Overall Score is computed by weighting the 4 statistical scores disproportionately (20% BB&HBP Score, 36% K Score, 22% LD&OFFB Score, 22% OFLD&OFFB Pull Score) and scaling the result away from 50 such that 10 points again amounts to 1 SD. Two SD better than average is outstanding in so much as such a pitcher is equaling or bettering 97% of their peers. One SD better than average is also good, as that pitcher is equaling or bettering 84% of their peers.
In the tables that follow, any of the 4 statistical scores that beat level-average by at least 1 SD are highlighted in green while those that trail level-average by at least 1 SD are in yellow. Numbers in parentheses indicate how many points by which the pitcher has improved versus his 2012 score, if available.
Of the 9 qualified hurlers, 5 rate rather well and 4 rate rather poorly. Foltynewicz’ 2013 scores make use of data from 2 leagues (Texas, California). Hallock’s 2013 scores make use of data from 3 leagues (Pacific Coast, California, Midwest); he lacks 2012 reference data due to an injury that cost him nearly the entire campaign.
Cruz has seen his BB&HBP Score drop by over 2 SD versus 2012 but is 1 to 2 SD better than level-average at the remaining 3 scores. Conversely, control is West’s forte as he currently rates 2 SD better than level-average at BB&HBP Score with small decreases observed in the other 3 scores over his 2-level jump from short-season A Tri-City in 2012. Foltynewicz bears a strong statistical similarity to the otherwise very dissimilar Cruz; while on one hand Foltynewicz’ control has lapsed such that his 2013 BB&HBP Score now rates nearly 2 SD below level-average, on the other he rates right around 1 SD better than level average at the other 3 skills and has improved noticeably in each of them versus 2012. Versus their same-handed level peers, Rodgers and Rollins are performing similarly to each other and solidly so at that.
There is relatively more reason for optimism here with the 7 qualified starters faring quite well as a group. The worst performer has a level-average 50 Overall Score. Of the 28 individual component scores, 10 (36%) are at least 1 SD above level-average (green) and just one is at least 1 SD below it (yellow). Given that Hauschild and Jankowski pitched sparingly in relief for 2012 Greeneville and that McCullers made only 4 starts for that club, the trio lack reference data for their 2013 scores.
Holmes continues to impress with his ability to strike out batters as a southpaw; the somewhat injury-limited 2013 data and the 73 Overall Score suggest that his strong 2012 Tri-City performance was no fluke. McCullers’ demonstrated ability to both miss bats (67 K Score) and avoid line drives and flyballs (63 LD&OFFB Score) surely places him in rather exclusive company as a 19-year-old toiling in full-season A ball. Hauschild’s 73 LD&OFFB Score is a testament to his extreme groundballing ways; for him to rate this well overall after only relieving for the Rookie-level 2012 Greeneville club is a pleasant development. This analysis paints Cain as a pitcher who is performing better than his traditional stats line would suggest; much like Holmes, Cain has been very difficult to pull aerially. Similar to Hauschild, Jankowski is more than holding his own as a tandem starter after pitching exclusively in short relief at Greenville in 2012. Versus his Tri-City performance in 2012, Velasquez has improved in his two weakest areas, LD&OFFB Score and OFLD&OFFB Pull Score, with no change in his average-for-level BB&HBP Score or his excellent K Score. Minor has had declines of nearly 2 SD apiece in the 2 batted ball stats versus his 2012 Greeneville starter data - thankfully those two values were rather good to begin with.
Bonus: Age Scores
Something else to consider when grading the pitcher’s performance is their age relative to level(s). In that vein, I have computed an Age Score to evaluate how many SD the player’s age is low/high relative to the mean of same-handed starters who pitched at the same level(s) in 2012. A 50 indicates level-average age, above 50 means that the player is young for their level, and any 10 points amounts to 1 SD.