Using xwOBA for Predictive Purposes in Pitchers

When it comes to predictive metrics, MLB's statcast has afforded the baseball community a wealth of data since its league-wide installation in 2015. One of the more intriguing metrics is expected wOBA, or xwOBA. This statistic gives us the expected wOBA for a player based on the hit probability of batted balls, calculated using exit velocity and launch angle. It takes fielding and batted-ball luck out of the equation, making it more predictive of a player's future performance than wOBA. This data can be found on the MLB's statcast website.

Unlike ERA predictors like FIP, xwOBA takes batted balls into consideration and may be more favorable to pitchers that rely on inducing weak contact compared to FIP. By comparing the difference between xwOBA and wOBA we can get more context for pitchers that may have overperformed or underperformed in 2017.

Rather than taking a broad brush approach this article will look closely at six pitchers, three that underperformed according to xwOBA and three that overperformed. This allows us to contextualize xwOBA with a player's performance since a large gap between xwOBA and wOBA does not necessarily mean a player will regress to predictive metrics. The pitchers analyzed in this article were taken from a sample that included all pitchers that threw at least 2000 pitches in 2017. The 2000 pitch barrier ensures that the pitcher made about 20 starts and threw over 100 innings. For context the average wOBA among this group was .323, and the average xwOBA was .313.

Editor's Note: Our friends at RTSports have best ball leagues with no in-season management. Just draft your team, and that's it! Use your phone for this casual draft by getting emails or texts when you're on the clock. Sign Up Now!

The Underachievers

Jameson Taillon, PIT - .341 wOBA, .303 xwOBA

Jameson Taillon had the eighth largest negative gap between his wOBA against and his xwOBA against among pitchers that threw at least 2000 pitches in 2017. Several underlying metrics suggest that Taillon suffered misfortune on the mound last season. The righty had nearly a one run gap between his 4.44 ERA and 3.48 FIP and had a .352 BABIP against last season. These statistics along with xwOBA suggest that Taillon is line for better outcomes in 2018.

Throughout his short career Taillon has allowed above average contact rates to batters. In 2017 he had an 82.2% contact rate, nearly 5% higher than the league average 77.5%. Allowing this much contact can inflate wOBA above xwOBA since more balls in play means more opportunities for hits regardless of the quality of contact. In 2017 Taillon had a stellar 22.1% soft contact rate, a number on par with pitchers like Kyle Hendricks and Stephen Strasburg. Coupled with his .352 BABIP against these numbers suggest that Taillon was unlucky regarding batted ball outcomes in 2017, and could see a drop in ERA, wOBA, and BAA next season. The Andrew McCutchen trade should help Taillon’s surface stats reflect his xwOBA and FIP. McCutchen had -16 defensive runs saved (DRS) in 2017, third worst among regular centerfielders. With improved defense and better luck Taillon should be an above league average pitcher going forward. With a 197 NFBC ADP Taillon could be a nice value this season.

Jeff Samardzija, SF - .314 wOBA, .294 xwOBA

Although the results have varied, Samardzija has been a reliable source of over 200 innings for five straight seasons. From a results standpoint, 2017 was not one of Samardzija’s best years. He posted a 4.42 ERA and allowed a career high 1.3 HR/9, however Several predictive metrics suggest that Samardzija may have been unlucky. He had a 3.60 xFIP and his xwOBA was 20 points lower than his wOBA. What’s most intriguing was his elite 6.41 K:BB, good for fourth best among qualified pitchers and putting him in the company of pitchers like Clayton Kershaw, Chris Sale, and Corey Kluber. Although Samardzija didn’t have the best results last season, there may be room for a mid-thirties resurgence.

The difference between Samardzija’s wOBA (.314) and xwOBA (.294) is the difference between average and good. His career high 20.7% soft contact rate over a full season is certainly a step in the right direction. He also allowed a slightly below average 30.1% hard contact rate and didn’t have an exorbitant flyball rate at 36.5%, which makes his 2017 case of Gopheritis even more frustrating. Samardzija’s xwOBA certainly thinks he should have had better results based on his batted ball data, but unless he finds a way to reduce home runs he will not regress to his xwOBA. His pronounced home/road splits make him nigh impossible to trust on the road, even in a neutral park. He allowed 30 home runs on the road in 2017 and had a 4.81 ERA. With a current NFBC ADP of 136 Samardzija is going too high for a 33 year-old with homer problems that can’t pitch on the road, regardless of how unlucky xwOBA and xFIP says he was. It’s probably worth taking a risk on breakout candidates like Zack Godley or Trevor Bauer around his ADP.

Marco Estrada, TOR - .338 wOBA, .299 xwOBA

Over the past few seasons Estrada had been the poster child for outperforming advanced metrics. After outperforming his xFIP by over one run between 2015-16 things fell apart in 2017. He posted a career high 4.98 ERA and 3.44 BB/9. Has Estrada’s contact-friendly pitching style finally caught up with him? With a 4.61 FIP, 5.09 xFIP, and 4.69 SIERA in 2017, most predictive metrics spell doom for the 34 year-old. One of the few metrics that offers optimism for Estrada is xwOBA. While Estrada posted an unsightly .338 wOBA against, he had a stellar .299 xwOBA. If Estrada’s 2017 wOBA regresses towards his xwOBA in 2018 he could see return to his 2015-16 value.

Since xwOBA incorporates hit probability by using launch angles and exit velocity, it stands to reason that Estrada has a low xwOBA. In 2017 Estrada had a 27.2% hard contact rate, 4.6% below league average. He also had a 21.4% soft contact rate, 2.5% higher than league average. Staying true to himself, Estrada led qualified pitchers with a 50.3% flyball rate and 16.6% infield flyball rate, a statistic where he has been routinely at or near the top of the leaderboard. This type of contact, infield popups and weakly hit flyballs, will drive down xwOBA since they have low hit probability. The biggest difference between 2015-16 and 2017 for Estrada was his BABIP against. Estrada had a .295 BABIP against in 2017, and while that number is around league average, it’s above his career .263 BABIP and far above his 2015 BABIP of .216 and 2016 BABIP of .234. The 39 point gap between his wOBA and xwOBA suggests that many of the hits surrendered by Estrada in 2017 were unlucky. Pitchers that live off inducing flyballs have always walked a tight rope, but if things broke the right way for Estrada in 2018 he’d make a good value at his current NFBC ADP of 308.

The Overachievers

Michael Fulmer, DET - .285 wOBA, .307 xwOBA

Michael Fulmer had the largest positive gap between his wOBA and his xwOBA in 2017. Several predictive analytics suggest that Fulmer overperformed in 2017. Despite his 3.83 ERA he had a 4.24 xFIP and .273 BABIP against to go along with the 22 point gap between his wOBA and xwOBA. Unlike metrics like xFIP, xwOBA takes batted ball profiles into consideration. This makes xwOBA a good metric to look at for pitchers like Fulmer, who perform well but are not big bat-missers. Even though Fulmer has a 6.84 career K/9, he allowed just a 30% hard contact rate and induced 10.2% infield fly balls in 2017. Both of those skills will help him keep his wOBA and xwOBA in a healthy range. It’s not as if the defense was helping Fulmer overachieve either. Collectively the Tigers had -62 DRS, second worst in the majors in front of the Mets. Even if Fulmer regressed to his .307 xwOBA that number was still five points below this sample’s average. Coming off ulnar transposition surgery Fulmer presents an injury risk heading into 2018, but don’t let his xFIP or low BABIP against scare you away. The kind of contact he allows suppresses hits, and xwOBA demonstrates this.

Sonny Gray, NYY - .295 wOBA, .307 xwOBA

Like Marco Estrada, Sonny Gray became known for outperforming predictive metrics like FIP, xFIP, and SIERA during his dominant 2014-15 seasons. Unlike Estrada, Gray beat predictive metrics by keeping the ball on the ground, with a career 53.8% groundball rate. He also severely limited his home runs, with a HR/9 lower than 0.75 in his first three seasons. Injuries derailed his 2016, but Gray bounced back nicely last year with a 3.55 ERA over 162.1 innings. Even with this recovery Gray did not return to the ace-level numbers he had before 2016. The twelve point positive gap between his xwOBA and wOBA along with his 3.90 FIP and 15.1% HR/FB suggest that regression could be coming for Gray in 2018.

Based on his ability to get groundballs, Gray should be able to maintain a low wOBA even if xwOBA thinks it should be higher. Part of the reason Gray’s xwOBA is higher than his wOBA was his below average 16.8% soft contact rate. If his soft contact rate stays that low Gray’s xwOBA will most likely always be higher than his wOBA. Even if Gray’s wOBA regressed to his xwOBA it would still be better than average and wouldn’t cause him to unravel. Considering his BABIP against was .269, around where it was in 2014-15, it seems as though Gray is getting the type of contact he needs to find success. To return to his ace form Gray needs to overcome his home run issues. When Gray’s sinker gets lifted it gets hit hard, which contributed to his 15.1% HR/FB rate. His move to Yankee Stadium from the forgiving Oakland Coliseum is worrisome, especially considering that nine of his 19 home runs allowed came in September while in pinstripes. If there is reason to think Gray will regress in 2018 it is not his xwOBA, which was still above the sample average. What should scare fantasy owners away is his inability to limit home runs coupled with a full season in second most homer friendly park per ESPN park factors.

Mike Clevinger, CLE - .299 wOBA, .311 xwOBA

Mike Clevinger garnered a lot of attention in 2017, primarily for his 10.13 K/9 and 3.11 ERA in 121.2. If Clevinger starts the season in Cleveland’s rotation he could be on the precipice of a breakout. The question remains, however, whether the 27 year-old that spent six years in the minors is legit. According to xwOBA, he overperformed his wOBA by twelve points, the sixth-largest gap among pitchers with at least 2000 pitches. Clevinger also had a 4.05 xFIP and 4.44 BB/9, meaning regression could be coming for Clevinger.

If Clevinger were to regress to his xwOBA, he would be around the average xwOBA for pitchers within this sample. Part of the reason that xwOBA expects regression is that Clevinger surrendered a 34.2% hard contact rate and a 24.1% line drive rate. Since the biggest factors in determining xwOBA are exit velocity and launch angle, having hard contact and line drive rates above league average will naturally increase xwOBA. Clevinger counteracts this type of contact with an elite strikeout rate, including an elite swinging strike rate of 12.4%. He walks a dangerous line with the amount of walks and hard contact he allows, and his xwOBA reflects this tenuous pitching style. Expect Clevinger’s ERA to increase in 2018 based on these numbers, but even if he regresses to his xwOBA his ERA won’t necessarily regress to his xFIP because of his dominant strikeout rate. He's not guaranteed a rotation spot, but if he is a starter going into 2018 he makes for a nice sleeper his current NFBC ADP of 218.