Statcast Data Primer For Diagnosing Player Performance
Is Yoan Moncada a “flyball revolution” guy? Has Francisco Cervelli raised his launch angle? What about Xander Bogaerts, has he become a different type of hitter? Did Trevor Bauer really cheat for an inning to prove a point? These are just a few of the questions that I’ve answered for myself or for others in recent days, and I use various views and aggregations of MLB’s Statcast data to dig deeper into the “Why?” part of these questions. In my first Statcast article, I gave a brief primer on using Statcast for evaluating players. This time, I’m going to put it to use.
What’s On Baseball Savant Player Pages (Statcast)
First, let’s familiarize with what you can get from the player pages at baseballsavant.mlb.com by looking at Cervelli’s page to try and answer the launch angle question I posed.
Sure enough, we can see that Cervelli has an average launch angle of 19.8° this year. At first blush, his weighted on-base average (wOBA) and expected weighted on-base average (xwOBA) look to be correlated with this change in approach at the plate. If we dive a little deeper on his Savant page, we can also see some rudimentary batted ball profile statistics.
Is he hitting more Flyballs (FB%)? Yes! Is he pulling the ball more (Pull%)? Not really. We can also see that he’s had more Barrels, has limited his weak contact (Weak%) and his tapped ground balls (Tapped%).
Now that we’ve baselined ourselves on Francisco Cervelli’s current batted ball the profile, the next logical questions you should be asking yourself are:
- How has Francisco Cervelli produced these symptoms?
- Is this truly a batted ball mix change or is it a small sample size outcome?
- Is this new profile sustainable, and can I use it to predict future outcomes for Cervelli?
For these deeper questions, I’ve found that I need to dig into the raw Statcast data and make some data visualizations of my own. Luckily for us data nerds, baseballsavant.mlb.com provides us with the ability to scrape all the data that’s available in the Statcast database to do our own analysis, which is useful in delving into the deeper explanations for the results we see.
In the case of Cervelli, I want to be better able to explain how he’s come to a much higher launch angle. Is it because he’s hitting a lot more balls right at 20°, or that he’s hit a lot more really high pop-ups that skews his average launch angle? A quick glance at his player page showed us his PU% (Pop Up) was up and his LD% (Line Drive) was down. This leads to further questions, like:
- Is Cervelli pulling the ball more in the air? If so, is he homering at a greater rate because of this?
- On what types of batted balls is he exceeding his expected outcomes?
- Yes, he’s barreling more balls, but is he barreling more fly balls? Liners? Ground Balls?
My Statcast Tool On Tableau Public
To answer these questions, there are a couple different data points I work with from baseballsavant.mlb.com. I use some combination of: Woba Value, Launch Speed, Launch Angle, Babip Value, Events and Estimated Woba Using Speedangle (Statcast’s version of expected weighted on-base average). Using these data points in combination with some of the more descriptive information given about each batted ball (Hc X, Hc Y) we can determine exactly where the ball was hit on the field, at what launch angles and at what type of velocity. I commonly use a chart that looks like this:
Recall that this is the “binning” and color-coding that we discussed last time. The first thing to know is that the red region is loosely correlated to LD%. I say loosely correlated because the way that Statcast, Baseball Info Solutions, Brooks Baseball, etc. have their own proprietary ways to categorize batted (and pitched) balls. For batted balls, they use combinations of launch angle, launch speed (commonly referred to as Exit Velocity), and hang time. For pitched balls, there are numerous classification systems, but I don’t believe anyone publishes their algorithms for this so it’s not entirely possible to replicate splits on specific pitch types (FB, CU, CH, SL, etc) exactly like you see them on Fangraphs (BIS provides theirs), Brooks Baseball.
Getting back to the red region on the chart, these are the batted balls that provide the highest BABIP and one of the highest wOBAs you’ll see. It’s singles-heavy, but you also get some doubles and triples in here as well. After the red region, there’s a grey area which corresponds to low flyballs, which are commonly referred to as “fliners” (when you take into consideration launch speed and hang time) because they are a cross between line drives and fly balls. It’s harder, but still possible to homer on these batted balls. The black region is primarily where you’ll see most home runs hit and the highest HR/BBE (Batted Ball Event). I’ve added J.D. Martinez as a comparison point in this chart. You can see he homers on 40-55% of these types of balls in a given season, but there’s still a range and some fluctuation.
I’ve now covered the highest value batted balls, but there are also some bins I’ve color-coded with pink values to describe their contribution to overall player wOBA as well. The bottom-most bin is “choppers”. These are ground balls that are hit straight downward. Maikel Franco loves to hit these. Generally speaking, since you’re rarely going to have a double or triple off a chopper, the wOBA value shown on the chart (E.g.: .227 for 2018 Cervelli) is synonymous with BABIP. For choppers, hitting them harder isn’t always better. A dribbler is more likely to produce a base hit here than a ball chopped at a higher speed to an infielder.
As you move upward through the chart, the next bin of batted balls are referred to as “Grounders” and are what you traditionally think of when thinking about ground balls hit at various velocities toward infielders. In this bin of batted balls, the harder you hit them the more that will get through the infield for hits. Again, while this chart shows wOBA, it’s pretty much synonymous with BABIP.
Lastly, on the more groundball side of the spectrum, we have what are referred to as “grinders”. These are balls that are a cross between a grounder and a line drive. Think of these as being launched directly at a fielder’s feet. Short hops, super low liners that hit before an infielder – that type of thing. I like to think of these as a “poor man’s line drive”. You can see in the wOBA values in the chart above that J.D. Martinez routinely runs a wOBA around .540 to .580 in this bin as compared to .750 to .850 in the red line drive bin.
That covers the ground ball to line drive side of the spectrum, so all that’s left is to cover the other types of fly balls. In the pink region just above the black homer-heavy bin, we have the other half of the “high drives”. These are commonly referred to as “floppers” – a cross between a flier (black bin) and a pop-up. Players with elite exit velocity and game power have the ability to hit more of these balls for home runs, though you can see I’ve color-coded them as producing approximately the same wOBA value as Grounders. Really strong players might homer on 10-20% of these types of fly balls. Gallo homers on 32% of these and Giancarlo Stanton 22%. You can see J.D. Martinez had a bit of an outlier season last year on this fly ball type, hitting 32% of those for home runs.
Finally, the last bin, which I sometimes refer to as “sky balls”, these are traditionally mostly pop-ups. If we were to apply a more sophisticated algorithm like those used on baseball savant or by baseball info solutions we could truly pull out just the infield pop-ups and separate them from outfield pop-ups, but for my purposes, knowing there is almost .000 wOBA value for all these batted balls I just keep them lumped together.
Cervelli’s Air Balls, New Found Power & Some Luck
Let’s bring this full circle back to Cervelli, and I’ll dial in on some points. First, let’s remove all his ground balls and just assess if he’s hitting more balls at the better launch angles.
No, not really. Nearly all of his gain in launch angle is coming from what I refer to as “sky balls” which have no additional wOBA value added to his profile. If I scroll my eyes through his optimal launch angle for home runs (black bins) I can see that he’s experiencing a HR/FB excursion on those batted balls. While he’s typically been at ~10% for his career (ranging from 0% to 16%) he’s at 30% this year. While he’s had a short-term spike in home runs (and therefore has his high wOBA in this bin), I’d chalk this up to being fluky absent any other evidence. However, you can quickly see that Cervelli is producing above career-average HR/BBE in all three of the batted ball types with which you can homer (13% Fliners; 30% Fliers; 17% Floppers).
One thing that we know can affect HR rates is the horizontal spray direction (Pull / Straight / Oppo) on fly balls specifically. On the baseball savant page you can see Cervelli’s total spray distribution (not just for balls hit in the air, or balls hit in the air at optimal launch angles for home runs).
Although I’m able to do that with the tool I built in Tableau. The blue color-coded chart shows the percentage of balls hit in the air at 95+ mph. Since we’re just looking at air balls, these are all what Statcast calls Barrels and are loosely synonymous with Fly Ball Barrel%. The green color-coded chart shows average exit velocity on these balls. NOTE – this average exit velocity is a pretty meaningless stat due to the fact that the average doesn’t properly describe anything to do with the overall quality of these batted balls, but alas, I show it. The bigger takeaway from the green chart is that I show the Spray% to the Pull / Center / Oppo fields. From this, we can discern that Cervelli is pulling the ball more in the air and if you were to hover on my Tableau Public tool you’d be able to see that Cervelli has homered on 14% of these air balls versus a career average of 20%. This tells me that he’s not “out of the norm” on these batted balls, even though he’s barreling a greater percentage of them. Centerfield is different though, as he’s homered on 22% of those air balls against a career average of 4%. The same goes for his opposite field air balls where he’s homered on 14% of those versus a career average of 3%.
With this information, I’m better prepared to answer the other questions we had on Cervelli and to make some judgment calls on his future value. The first thing I do in these situations where I have new found HR/FB spikes is Google for swing change articles. I did this on Cervelli and found one, but it’s behind a paywall. What we do know about Cervelli is he’s not hitting a greater percentage of his balls at these good angles, but he is barreling more. His .743 wOBA to the opposite field against a career average (Statcast Era) of .420 seems like an outlier even if he’s made swing changes. I’m not likely willing to buy that he’s going to be as good as JD Martinez to the opposite field in the air without hitting a greater percentage of his balls at optimal launch angles or at elite exit velocity more frequently. Similarly, I think the HR/BBE spike to Centerfield is an aberration as well, and he’ll slowly come back to down from .722 to maybe settle in the .500-.600 range. The pull field power looks to be for real, so he should continue seeing improved performance from that change in his game.
Cervelli’s Ground Balls & Luck
As a final display of the types of information I look at for players, I’ll quickly show BABIP on ground balls together, as well as average home run distance (a favorite stat to quote for @TheGreenMagnus on our podcast) and median launch angle.
Cervelli is certainly also hitting his ground balls harder, but his .417 BABIP on those for a non-elite runner is most certainly going to regress toward his career average (and drain his batting average and on-base percentage) as we move forward throughout May. If you’re currently riding him as a hot hand you could opt to let his AVG/OBP tumble before you go back in for more of his power if you have another catcher you can play.
If you made it through this 2,000-word monstrosity and want to reach out to me with any questions, please feel free to find me on Twitter @JimMelichar7.