Using Sabermetrics for Fantasy Baseball Part 14 - Statcast For Pitchers

Previously, we looked at Barrels, a stat combining exit velocity and launch angle to measure how often a batter makes quality hard contact. As much as batters want to hit a Barrel every time, pitchers want to avoid them at all costs. Yet there is some evidence that pitchers do not have the same influence over Barrels as a batter does.

While Aaron Judge led all of baseball last year with 86 Barrels, Rick Porcello led all pitchers by coughing up 52. Neither performance was an outlier, so it seems to take fewer Barrels to lead pitchers in Barrels given up than it does to lead hitters in Barrels hit. This fits well with DIPS theory, which states that batters can do more to influence batted balls than pitchers can.

It's also not fantasy-relevant, as Porcello's 2016 was a fluke by any predictive metric. Ian Kennedy allowed the second most Barrels with 51, but fantasy owners don't care about him. Matt Moore came in third with 48, but he hasn't been fantasy-relevant in years. Fourth place Ricky Nolasco (46) is bad, and four arms tied for fifth with 45 Barrels allowed. Kevin Gausman, Jason Hammel, and Ariel Miranda are blah, but Gerrit Cole is interesting. Let's start with him.

Editor's Note: For a limited time only - get a full-season NFL Premium Pass for 50% off, and you'll also get MLB Premium for free! Our Preseason Draft Kit, In-Season lineup tools and over 150 days of Premium DFS research. Sign Up Now!

How to Interpret Batted Ball Statistics

Cole allowed hard contact in 2017, but nothing in his history suggested that he would before the season started. In 2016, he allowed only 11 Barrels all season. In 2015, he allowed 26. Cole's high number of Barrels allowed may partially explain why he disappointed his fantasy owners last year, but his Barrels allowed look like they came out of a random number generator. There's nothing predictive here.

The rate stat, Brls/BBE, might seem like a better option. Jered Weaver tied for the league lead in rate of Barrels allowed with 11.8%, and he's obviously terrible. The person he tied with was Craig Kimbrel, one of the best relievers in baseball.

The Barrels hardly hurt Kimbrel's final stat line, as he posted an elite 1.43 ERA (1.50 xFIP) with 35 saves last season. Kimbrel had previously been great by Brls/BBE, posting a 5.8% mark in 2016 and 4.7% in 2015, so nothing in his track record should have raised a red flag. Indeed, there's no need for a red flag even in retrospect.

Maybe we need to simplify this and just use average airborne exit velocity? Cesar Valdez (97.2 mph), Sam Dyson (96.1 mph), Brett Anderson (95.5 mph), Scott Alexander (95.4 mph), and Chasen Shreve (95.3 mph) top this list, but none of them are on the fantasy radar. Nate Karns kind of is (also 95.3 mph), but he didn't really struggle with average airborne exit velocity in 2016 (93.3 mph) or 2015 (92.6). Again, there is nothing predictive about these Statcast metrics.

Last year's version of this article cited Chris Archer and Justin Verlander as case studies for the value of these metrics. Archer under-performed his peripheral stats in 2016, posting a 4.02 ERA against a 3.41 xFIP. I speculated that the quality of contact he allowed (46 Barrels, 8.4% Brls/BBE) may be the reason why. Last season proved this to not be the case, as Archer improved both Statcast metrics (29 Barrels, 5.4% Brls/BBE) while still posting an ERA (4.07) above his xFIP (3.35). Archer has some kind of issue, but Statcast metrics are not the way to quantify it.

Likewise, Verlander's 2016 xFIP (3.78) was considerably above his ERA (3.04). He allowed a ton of Barrels (45) and a high rate of Brls/BBE (7.7%), metrics I used to forecast regression. Nothing changed in 2017, as he again allowed a ton of Barrels (40, 7.2% rate of Brls/BBE) while posting an ERA better than his xFIP (3.36 vs. 4.17). The statistics seem to have been proven worthless in both cases.

Ultimately, Statcast metrics such as Barrels and average airborne exit velocity should probably just be ignored for pitcher analysis. These metrics are great for evaluating batters, but I can't get anything out of them for pitchers even with the benefit of hindsight.

Conclusion

That conclusion may make this seem like a worthless article, but it isn't. Every fantasy analyst uses contact quality to credit or penalize pitchers, either through the Statcast numbers above or an approximation such as Hard%. This type of analysis may explain a pitcher's performance after the fact, but it seems to have zero predictive value. Therefore, there may be a competitive advantage to be gained by ignoring this type of analysis completely. Score it as a win for DIPS theory.