One of the few good stories that have risen from the ashes of this horrible April for the Toronto Blue Jays has been the continued success of Marco Estrada. So while not much good is going on, let us take a look back at the numbers of Marco “the enigma” Estrada using some neat new tools, and maybe solve a mystery in the process.
In two full seasons since coming to the Blue Jays in November of 2014, Marco Estrada’s results on the mound have been nothing short of elite. By ERA, Estrada’s mark of 3.30 over the past two seasons is ranked 20th among major league qualifiers, nestled between Jose Quintana and Corey Kluber. He’s also been dominant in the playoffs, pitching to a 2.18 ERA in 41.2 innings and saving the 2015 Blue Jays season on a couple of different occasions. All of that is fantastic in its own right, but it’s made all the more amazing when you consider that he’s done this in the American League East with a fastball that averages under 89 mph.
The borderline paradoxical difference between Estrada’s results and his style of pitching has driven the conversation around him and explains why he’s been so underrated by the league. The reluctance to buy in to what Estrada has done has stemmed mainly from his low velocity and some pedestrian peripherals. In 2015, Marco Estrada’s ERA was ranked 17th out of 73 qualifiers, but his K% and BB% of 18.1% and 7.6%, respectively, were below average. As a result, Estrada wildly outpitched his FIP (a stat that’s similar to ERA but only includes walks, strikeouts, and homeruns while excluding batted balls in play) and posted the lowest E-F (FIP subtracted from ERA) in the league at -1.27. In 2016, most of these trends remained and Estrada outpitched his FIP again. He finished with an E-F of -0.67, 8th lowest in the league.
How has Estrada allowed so few runs with such mediocre strikeout and walk percentages? The answer, of course, is BABIP. The batting average on balls put in play against Estrada was .216 in 2015 and .234 in 2016. He led the league for both years in this category. This is undoubtedly responsible for Estrada’s great run prevention over the past couple of years, but there is much confusion and disagreement around how to interpret this.
Pitcher success driven by low BABIP is not seen as valuable compared to success resulting from good strikeout and walk percentages. BABIP is viewed by many as simply a result of random luck and the quality of the defense behind a given pitcher, and therefore not as important when evaluating pitchers as other peripherals.
However, some pitchers manage to be quite good at continually suppressing their BABIP. Max Scherzer, Jake Arrieta, and Hector Santiago all joined Marco Estrada as having BABIPs amongst the lowest thirteen in both 2015 and 2016. Tony Blengino of FanGraphs has written that Estrada has been elite at limiting grounders and line drives while inducing pop-ups at an incredibly high frequency. He argues that this batted ball profile is ideal for minimizing the likelihood of batted balls from turning into hits.
Therefore, we know that Estrada’s low BABIP is essential to his success and we know that BABIP is a mix of luck and defense, which Estrada can’t control, and the quality/type of contact that he surrenders, which he can control to some uncertain degree, but we don’t know the relative importance of these factors in suppressing his BABIP. For prediction purposes, it would be valuable to know how much of Estrada’s 2016 BABIP suppression was his own doing.
To do this, the relatively new expected batting average (xBA) feature on Baseball Savant will be utilized in a manner similar to how Craig Edwards used it in an earlier piece over at FanGraphs. The xBA feature predicts the expected batting average of batted balls in play based on their exit velocity and launch angle. Looking at this will allow us to directly hone in on the quality of contact allowed by Estrada without luck and defence getting in the way.
The first step is to take all the balls in play (BIP), league wide, from the 2016 season with expected batting averages and to separate them into discrete intervals. Overall, around 105 000 of 120 000 balls in play had xBAs calculated for them on baseball savant. Displayed below are the numbers of balls in play as well as BABIPs for each group.
These results follow what one would expect to see. Balls in the below .200 group rarely fell for hits and balls in the above .600 range fell for hits over three quarters of the time. The “no xBA data” group will be ignored going forward as will the sacrifice fly group since sac flies aren’t counted as BIP on baseball savant for some reason. Next, I added the same columns but only for balls in play allowed by Marco Estrada.
Interestingly, Estrada had lower BABIPs in every group but one compared to the league. This means that even after accounting for quality of contact, Estrada still seems to have allowed fewer hits than the average pitcher would have on batted balls in all xBA groups, except for those batted balls with an xBA between .300 and .400. So, it appears that there must have been some good luck and/or good defense on Estrada’s side last season.
Given this, we now want to know how many extra outs Estrada gained for every xBA group and in total.
Estrada gained approximately 22 outs compared to what he should have gotten with league average luck and defence. These gained outs come from all xBA groups except for the .300-.400 group where he was worse than average by about 3 outs. The group where he gained the most outs was in the .600-1.000 group. In this group, 46 of 73 balls in play landed for hits, while the average pitcher would have had 56 of those 73 fall behind them.
Some of these gained outs in the .600 and above group were probably a result of good fortune. However, anyone who follows the Blue Jays would probably first look to their defensive star in centre field when thinking about why scorched balls found leather instead of grass (or turf).
Kevin Pillar, who became the everyday Blue Jays centre fielder around the same time that Estrada entered the Jays rotation, has been an elite defensive outfielder in the league over the past two years. In 2016, he posted an eye popping 21 DRS and 26.3 UZR/150.
Therefore, using Baseball Savant, a spray chart was generated looking at the 27 of 73 balls in play against Estrada that had an expected batting average above .600 that resulted in outs.
Clearly, most of the outs on these hard-hit balls came in the outfield. Looking deeper, an amazing 11 of the 27 outs came via Kevin Pillar’s magic in centerfield (Pillar’s put outs are circled in red). The most memorable of these was the one with the red arrow pointing to it. I recommend that you check it out.
It appears as though Kevin Pillar’s incredible defense in centerfield and the Jays great team defence in general (they were 8th in team DRS in 2016) bought Marco Estrada a lot of extra outs on balls with a .600 or above expected batting average. It’s hard to separate good defense and luck when explaining BABIPs, but its probably fair to say that good defense played at least an equal role as good fortune in this case.
Getting back to Estrada’s’ BABIP, if we subtract those 22 gained outs, Estrada’s BABIP jumps from .234 to .281. That 47-point jump takes Estrada from having the lowest BABIP in the league to being tied with Chris Tillman for the 27th lowest BABIP among qualifiers. That is still good but not great.
The difference between the league average BABIP of .300 and Estrada’s BABIP of .234 was .066 (or 66 points). If 47 of those points came from factors outside of his control including luck and defense, then the unaccounted for 19 points of difference must have come from Estrada’s often-discussed ability to manage contact. Using our xBA groups, Estrada’s management of contact can be visualized by looking at the proportion of balls in play against him in each group and comparing to the league.
Estrada had nearly 10% more batted balls in the below .200 group compared to the average pitcher (51.8% vs. 42.3%). He also allowed a smaller proportion of batted balls in the .600 and higher group than the league (18.6% vs. 21.6%). Clearly, Estrada was able to induce more poor contact and consequently less hard contact than the average pitcher. It is this contact management ability of Estrada’s that is responsible for depressing his BABIP by that final 19-points below league average.
Overall, the relative importance of the different factors in lowering Estrada’s 2016 BABIP can be summed up like this:
Overall, this xBA based analysis of Estrada demonstrates that his 2016 BABIP suppression (below the league average) was, surprisingly, more so a product of luck and defense than contact management skill. Of course, these numbers aren’t perfect because many batted balls had no xBA, and organizing data into buckets can cause statistical noise. Also, this isn’t necessarily bad news for Estrada. The 2017 Jays defense isn’t too different than it was in 2016. Most importantly, Pillar is still patrolling centerfield. Therefore, if Estrada can manage contact similarly to how he did last year and if the Jays defense does its thing again, then it’s hard to see his BABIP rising more than 20 or so points even if his luck takes a hit, and that would still put his BABIP down amongst the league leaders once again. After five starts in 2017, his BABIP is up at .289 and he’s got a sparkling ERA of 2.70, but the sample size is too small to draw any meaningful conclusions regarding BABIP.
So, it seems like we may be in store for another fun season of Estrada putting up zeros and embarrassing opposing batters, but why we’re marvelling at him, let’s not forget to give the eight guys around him some appreciation too, because Estrada’s success really is a team effort.