Now that we’ve passed the quarter-pole of the 2019 season, we are starting to be able to deep dive into the underlying statistics with more confidence. A quick way to see which pitchers we could argue have been unlucky is to compare their xwOBA (derived from Stat Cast batted ball data) against their BABIP.  Because xwOBA is based on the quality of the contact that a pitcher gives up, it would stand to reason that a low xwOBA should correlate highly with a low BABIP. Therefore, a pitcher who has produced a low xwOBA (expected offensive performance against them based on the quality/velocity/angle of batted ball events) but a high BABIP (batting average on balls in play) would suggest that they are pitching very well but due to ‘luck’ factors (such as defensive positioning) they have not gotten outs where expected. Conversely, a pitcher with a high xwOBA but a low BABIP would reasonably be deemed to have been lucky thus far.

Using season-to-date data on qualified Starting Pitchers, I plotted the xwOBA (as of May 18^th) on the y-axis and BABIP on the x-axis. Pitchers who have suppressed offense the most (ie have the lowest expected wOBA against) would be near the top of the graph and those with the highest xwOBA against would be near the lower part of the graph. Similarly, those pitchers who have given up the highest BABIP (batting average on balls in play) are on the right side of the graph and those with the lowest BABIPs would be in the far left of the graph. Those pitchers who have theoretically suppressed the hitting the most (low xwOBA) but for whatever reason have given up high BABIPs would be in the top right quadrant (unlucky). Pitchers in this area should have actually had better performance than they have had. Conversely, those pitchers in the lower left quadrant have actually likely over-performed (because they have had low BABIPs despite having given up high quality batted ball events) and are deemed to be lucky.

To help with the visualization, each ‘grid line’ represents one standard deviation. In other words, the average BABIP of the qualified pitchers was 0.287 (and is in the middle of the graph) and one standard deviation of BABIP (the x-axis) is 0.037; similarly, the average xwOBA (y-axis) is 0.314 (with a standard deviation of 0.034)

Beside the pitcher’s name, I have included their season ERA to date (as a quick metric of how they have done so far this year). For context, the average ERA is 3.84 with a standard deviation of 1.11. In other words, a pitcher who is one standard deviation better than average would have an ERA of 3.84-1.11 or 2.73.

Looking in the top right quadrant, we can see that the pitchers who are the ‘furthest’ away from the center-point (and have high BABIPs despite suppressing offense) are:

• Blake Snell
• Noah Syndergaard
• Gerrit Cole
• Eduardo Rodriguez
• Max Scherzer
• Zack Wheeler
• Stephen Strasburg

The way to interpret the graph is to see that Blake Snell has an xwOBA more than 2 standard deviations better than league average (98^th percentile) and has an approximate league average BABIP but despite this, his ERA is barely above league average at 3.31. All things being equal, we would expect his ERA “should” be in the top 98^th percentile or around 1.90. If you look at who else is clustered at the same xwOBA as Snell in the graph, we have Tyler Glasnow (ERA of 1.86) and Luis Castillo (ERA of 1.90). Interestingly, Stephen Strasburg is also there with an ERA of 3.32.

Moving in toward the center point, we encounter Gerrit Cole (xwOBA in the top 5% but ERA barely above average at 3.56), Noah Syndergaard (top 10% xwOBA, one of the top 15% highest BABIPs against) and a corresponding 4.84 ERA. Worried owners of Thor should at least have comfort in the fact that his ERA “should” be in the low 3’s. Whether this is predictive will become clear as the season progresses, but at least be aware that his ‘true’ performance is actually still of an elite pitcher.

Similarly, we see that pitchers with ERAs far higher than their xwOBA suggests are Eduardo Rodriguez who has a better-than-league-average xwOBA but due to his high BABIP has ended up with a sky-high 4.84 ERA, Zack Wheeler (ERA of 4.85) and to a certain extent Max Scherzer (ERA of 3.72).

The luckiest Pitchers:

Taking the flipside and looking the bottom left quadrant, we find the pitchers with the highest xwOBA against (theoretically providing the least effective offensive suppression based on quality of batted ball data) but who have for whatever reason had the lowest BABIPs are:

• Yonny Chirinos
• Trevor Bauer
• Shane Bieber
• Andrew Cashner

The two biggest names that jump out are the Cleveland tandem of Trevor Bauer (ERA of 3.76) and Shane Bieber (ERA of 3.81). I think most owners would dismiss their “poor” performance (relative to their pre-season expectation) as a minor speedbump – and would expect them to bounce back and regress back to their expected talent level. Unfortunately, that narrative may actually not be true at all. The data shows that Bauer (league average xwOBA but a BABIP that is more than 1 standard deviation lower than average) and especially Bieber (xwOBA that is in the bottom (worst) 12% in the league and BABIP that is one of the lowest 12% in the league) have arguably been lucky this year. As we showed in an earlier article, xwOBA correlates highly with ERA and crudely an xwOBA such as Bieber’s “should” correlate to an expected ERA of 4.95…much higher than his current 3.81.

Looking for the lowest ERA of players in the lower left quadrant, we find Yonny Chirinos. The Ray pitcher who usually takes the bulk innings after an opener has performed very well thus far in 2019 with an ERA of 3.26. Owners (such as myself) have begrudgingly accepted his K9 is in the 6’s because his great rate stats have been extremely helpful in this current pitching environment. Unfortunately, this may all soon come crashing down as the graph shows that although he has a league average xwOBA, his BABIP (0.211) is 2 standard deviations lower than the league average of 0.287. Grasping at straws, a Chirinos owner might convince themselves that the Rays are at the forefront of analytics and may have insights into how he can continue to suppress BABIP – or that, in fact, he has not been lucky at all to have such a low BABIP but it is a product of the Tampa Bay defensive shifts. Rationally though, I think regression (to an ERA in the low 4’s) should be expected.

Andrew Cashner is an interesting name, not because he is a relevant pitcher, but because his ERA of 4.10 might have fantasy owners wondering if he actually should be relevant. Long story short, the answer is a resounding no. His xwOBA is one of the worst 5% in the league and his ERA should not be anywhere as close to league average as it is.