Posted: 9:00 a.m. Thursday, Aug. 22, 2013
Let's dig into two sabermetric topics today. We will ask some questions about the rookie pitchers in the Astros rotation...like can the successes be sustained, and can the bad performances be improved? We may not have the answers, but at least we can look more deeply at the indicators.
1. THE ROOKIE STARTING PITCHERS
The Astros have three rookie starting pitchers in the rotation: Brad Peacock, Brett Olberholtzer, and Jarret Cosart. They join two other young starting pitchers, Jordan Lyles and Dallas Keuchel. Lyles and Keuchel have posted poor and mediocre ERAs, respectively. Olberholtzer and Cosart have begun their ML careers as starting pitchers in amazingly good fashion, with ERAs below 2. Peacock has an ERA over 5, but since his recall from Oklahoma City, he has a 1.86 ERA.
Most likely, at some point in the future, Olberholzer and Cosart will face regression (upward) in their ERA. Similarly, .Lyles and Keuchel have pitched better than their ERA, and are likely candidates for downward regression in their ERA.. This pattern is suggested by the pitchers' Fielding Independent Pitching (FIP), which is higher than the ERA for Cosart and Olberholtzer and less than the ERA for Lyles and Keuchel. The direction of Peacock's future ERA is less obvious, but most likely his true skill would result in a ERA better than his experience early this year, but not as good as he has pitched in the last month (Peacock has a .174 BABIP in August!).
All of this doesn't tell how much regression to expect. And the sample size for the rookie pitchers (26 - 45 IP) is so small, that we can't reach any accurate conclusion about their sustainable level of performance. But that doesn't prevent us from looking more closely at some of the factors behind their current performance.
Pitchers' Batting Average on Balls in Play (BABIP) is relatively volatile from season-to-season. But we can calculate expected BABIP (xBABIP) which tells us the probable extent of regression based on the types of batted balls allowed by the pitcher. This Talking Sabermetrics article last November discusses the x-BABIP formula. Strictly speaking, x-BABIP doesn't predict a pitcher's future BABIP; instead, it shows the expected BABIP if the pitcher's batted ball percentages continued in the future. The pitchers' 2013 BABIP and x-BABIP are shown below:
2013 BABIP and x-BABIP
Starter Avg. (AL)
As shown above, Cosart has the biggest spread between his actual and expected BABIP. If Cosart continues the same batted ball profile, he is the most significant candidate for regression in number of hits. Olberholtzer has a low BABIP, but his batted ball profile to this point results in an expected BABIP close to his actual BABIP. Keuchel and Lyles both experienced high BABIP, but their x-BABIP indicates the potential for improvement. Peacock is a candidate for moderation upward regression in BABIP.
HR/Fly ball is another statistic that tends to normalize or regress toward average over time. However, that doesn't rule out the possibility that some pitchers can suppress HR/Fly balls below average, while others are prone to give up more than the average rate of home runs. For example, I used Fangraphs to compile a list of starting pitcher career rates for HR/flyball since 1995. This resulted in an average HR/Fly rate of 10.6%. However, 1 standard deviation encompasses a range of HR/fly rates from 6.29% to 14.89%. There are pitchers (Matt Cain is an example) that have pitched more than 1,000 innings who have sustained HR/fly rates well below league average.
|Starter Avg. (AL)||10.9%|
Based on the chart, above, Cosart and Olberholtzer have HR/Fly rates which probably are unsustainable. But that doesn't necessarily mean we should expect regression fully to the league average rate. It's possible that they can sustain HR/fly rates below league average, but we don't have enough data to reach a conclusion. Cosart has a history in the minor leagues of HR/fly rates substantially below league average. Olberholtzer does not. Lyles and Peacock have experienced HR/Fly rates which are unsustainably high. Regression in the future is likely to improve their HR/fly rate.
Pitch f/x plate discipline stats constitute another indicator. A below average zone contact rate is normally associated with excellent pitching skill. Outside the zone swing rates and zone percentage are also useful in gauging the pitcher's ability. Olberholtzer is the only one in the group who stands out as above average for all three measures relative to the league. We can compare and rank Olberholtzer against the 34 other rookie starting pitchers in the major leagues. Olberholzer is 6th best at putting the ball in the zone (Zone %), 10th best at avoiding contact on his pitches in the zone (Zone Contact%), and 3d best at inducing hitters to swing outside the zone.
On the other hand, Cosart's ranking on plate discipline results among rookie starting pitchers are not strong from a surface review.. Cosart and Peacock rank worst at inducing swings outside the zone. Cosart also has the worst rate of pitching in the zone (38%). Cosart also has the highest ranking in zone contact percent (97.3%). The latter result is surprising, given Cosart's overall success. Perhaps Cosart is the unusual instance of a pitcher who induces a lot of weak contact on his pitches in the zone.
Although both Cosart and Oberholtzer have gotten off to similar great beginnings to their careers as starters in the majors, Oberholtzer's performance seems more nearly sustainable. Of course, the sample size is so small that even this conclusion is largely speculative. If Cosart undergoes extensive regression in his performance, an ERA equivalent to his current FIP,(3.79) would still represent a successful career.