Which White Sox players had luck on their side in 2017?

The concept of BABIP (Batting Average on Balls in Play) has established a foothold in common baseball discussion as a rough proxy for the amount of “luck” inherent in a hitter or pitcher’s results during the course of a season. Voros McCracken (of Moneyball fame) first theorized that a pitcher doesn’t have all that much control over what happens once a ball is put into play and that the percentage of balls in play that fell for hits didn’t correlate much from year to year for a particular pitcher. There’s similar year-to-year variance for hitters that can be mostly explained by luck. Hence, many will point to a pitcher who’s experiencing a season with a higher BABIP-against and say that this pitcher has been unlucky. Similarly, when someone like Avisail Garcia posts a BABIP of around .392 (far in excess of his career average of .340), the quick reaction is that Garcia has had a fairly lucky year.

McCracken’s theory was groundbreaking, but as batted ball tracking has improved over time, the understanding of the value of a ball in play has evolved. Some of that evolution is actually rather intuitive at its core. For instance, if a pitcher allows eight consecutive line drives in play and five of them fall for hits, the BABIP-allowed will be .625. That sounds pretty high, but the pitcher hasn’t experienced much bad luck here — he’s allowing batters to hit line drives! Using proxy data from 2014, we can see that line drives go for hits about 69% of the time, grounders go for hits about 24% of the time, and fly balls* are hits about 21% of the time. Clearly, the batted ball profile for either a pitcher or a hitter can have a big say in what that player’s BABIP (or BABIP allowed) should typically look like.

*fly balls include home runs, so the batting average on fly balls is likely higher in the homer-crazy 2017 than it was in 2014.

Within the past few years, Statcast has gone live and now we have a much more complete database of information regarding batted balls. They’re no longer just classified into ground ball, fly ball, and line drive or “Hard”, “Medium” or “Soft” contact. Now, we have specific data on the launch angle and exit velocity of each batted ball. Those two numbers encompass nearly all of the information required for us to determine the hitter’s quality-of-contact. It’s true that a hitter can aim to a specific side, but there’s no hitter than can precisely aim a projectile traveling 90 mph from 60 feet 6 inches through the hole between short and third or perfectly between two outfielders. All the hitter can do is make good contact and maybe target a certain general area of the field as the situation dictates.

Therefore, almost all of the “skill” in the showdown between a pitcher and a hitter can be boiled down to four things: 1) strikeouts, 2) walks, 3) exit velocity, and 4) launch angle. The rest of what makes up the results (such as defensive positioning*, defensive skill, the precise trajectory of the ball, and park shape) are basically out of the control of the pitcher or hitter. Using Statcast data, we can determine the probability that a particular ball hit at a certain angle and velocity will fall for a hit. We can also determine how likely it is that it will go for a single, double, triple, or home run.

*excluding radical shifts that are a direct reaction to a hitter’s extreme batted ball profiles

Armed with this information, we can calculate an “expected weighted on-base average”, which the Statcast glossary abbreviates as “xwOBA”. By comparing a player’s xwOBA to their actual weighted on-base average, wOBA, we can determine which players’ results have been better or worse than they “should” be given the quality of contact that they’re making. There’s three important things to note, however:

1. The speed of the player can have a big say in whether some batted balls will go for infield singles, doubles instead of singles, or triples instead of doubles. The xwOBA stat doesn’t account for player speed, so all else equal, we’d expect a faster player to outperform their xwOBA.
2. wOBA isn’t park-adjusted, but xwOBA effectively is because it’s only taking account how hard the ball’s being hit and at what angle. Therefore, players in hitters’ parks are more likely to have a wOBA that outperforms their xwOBA.
3. The 2017 data has a bias toward wOBA exceeding xwOBA. I have reached out to Daren Willman on this and have yet to receive a reply, but I strongly believe that the “juiced ball theory” is the culprit. If the balls being used in 2017 are less susceptible to air resistance, they’re more likely to become home runs than xwOBA thinks, as xwOBA is based off of comparable batted ball data since 2015.

Here’s a look at the difference between actual and expected wOBA for the 2017 White Sox hitters. This list is sorted from “most unlucky” at the top to “luckiest” at the bottom.

Here’s some key takeaways from this chart to highlight:

No. 1: Good news for Avisail Garcia

It’s widely believed that Garcia’s .392 BABIP has fueled his stellar season. That’s true to a degree, but xwOBA doesn’t see him falling off all that hard. The .353 figure above still suggests he’s a very good hitter even without the fortuitous bounces. With his vastly improved defense in right field, Garcia looks the part of a well above-average player even when his luck comes back to earth.

No. 2: Not such a favorable outlook for Nicky Delmonico

Delmonico, on the other hand, has just a .277 BABIP, so not many were pointing to his 2017 results as particularly fortunate. He earned a big chunk of his wOBA with favorable walk and strikeout rates, but his contact profile suggests he’s due for some regression. Per FanGraphs, his 27.3% soft contact rate would have been the highest in the major leagues among qualified hitters. That includes his three strategic bunts for hits (which xwOBA actually dings him for), so it’s not a perfect metric, but Delmonico will need to start squaring up pitches better to maintain his results. The good news is that even when regressed to expectations, his 2017 performance at the plate was probably still slightly better than league average.

No. 3: Cody Asche got hosed, man

He may have been abysmal in his brief White Sox tenure, but xwOBA says that he should have merely been pretty bad!

Now, let’s take a look at the White Sox pitchers. Once again, the list is sorted such that the “most unlucky” pitchers are at the top and the “luckiest” are at the bottom. Since we’re now talking about wOBA allowed, the “lucky” pitchers now have negative numbers in the “difference” column.

Here’s some things that stuck out to me about this chart:

No 1. Lucas Giolito might fight regression well

The elephant in the room is Giolito’s .189 BABIP, which is impossible to maintain even if you’re great at controlling opponent contact. The above suggests that Giolito wasn’t being hit all that hard when the ball was being put into play. It’s a small sample and the league will likely have more information on him next season, but at least we can say that his stuff was difficult to square up in his first stint in the junior circuit.

No 2. Reynaldo Lopez was fine

This is where I start to have some skepticism about the accuracy of Deserved Run Average for some pitchers, particularly in smaller samples. Most metrics, including ERA, FIP, and your standard peripheral statistics would indicate that Lopez’ performance was about on-par with a serviceable back-end starter. Baseball Prospectus has Lopez at a 7.34 DRA and -0.9 WARP, both of which suggest that he was an unmitigated disaster. Here, it’s indicated that Lopez allowed a league average wOBA at .321 and xwOBA is saying that this is a fair indicator of how well the league hit him.

No 3. Jake Petricka’s bad luck might cost him some money

Of the pitchers above, Petricka under-performed his xwOBA the most. That’s a problem for him, because he’s going to be up for arbitration a second time and his 2017 results and injury history make a pretty good case to not guarantee him $1 million or so. His .322 xwOBA isn’t good by any means for a reliever, but if his results matched, he would have been serviceable enough to all but guarantee him a spot in next year’s bullpen as an experienced arm on a staff starved for those. Petricka’s biggest problem was that he allowed six home runs, but four of them had no better than a 60% chance of becoming homers based on contact quality (and two were really cheap ones).

No 4. Regression cannot save you, Dylan Covey

Yeah...