Carlos Gonzalez is not an upgrade over Matt Holliday: A batted ball analysis

Matt Holliday has long been an under-appreciated and underrated Cardinal. While I've only been a writer here for the last for eight or nine months, I've been a member here for over three years. In that time, Holliday has seemed to be much more properly rated here than other bastions of Cardinals fandom. But even here, there's been a lot of talk in the comments lately about how Holliday may have finally started declining hard.

It's not all that hard to see on the surface. Holliday's 110 wRC+ currently is the lowest since his rookie year where he posted a 104 wRC+. Of course, Holliday's BABIP of .247 is much lower than his career total of .334, and much lower than his previous career low in BABIP of .298. Still, critics maintain that at his age, and an increase in both FB% and Pull% means it makes sense that his BABIP is lower. Those are all good reasons for why a hitter's BABIP may decline, but not enough to make up for such a large drop-off.

Some have went so far as to suggest the team should trade for Carlos Gonzalez. "CarGo" is currently sporting a 126 wRC+ playing his home games in the most unique MLB stadium on earth (Coors Field). He is earning $17M this year, so would be owed a little less than $6M for the year the day after the trade deadline, and is set to earn $20M in 2017, the final year of his contract. The idea is that the team could upgrade with Gonzalez now, and decline Holliday's option next year with Gonzalez in the fold. Here's a look at Holliday and Gonzalez's numbers:

Holliday walks slightly more, strikes out less, has hit for a little less power (before adjusting for the fact that Gonzalez's home park is on the moon), and the biggest difference: Gonzalez has a staggering lead of 123 points in BABIP.

There's also the fact that he plays at Coors. Looking at Fangraphs' Park Factors, you see that Coors isn't only the most homer-friendly park, it's also the most single-friendly and triple-friendly. In doubles, Coors is only beat by Fenway Park, thanks to the Green monster.

That's because the ball carries so well there that they keep the outfield dimensions rather deep just to lmit homers at their current league-leading rate. However, that also means more ground for the outfielders to cover, which means more balls fall in. Compare that to the Reds' Great American Ballpark or the Brewers' Miller Park, both of which are not above-average at anything except homers.

As you probably know, BABIP (batting average on balls in play) takes an awfully large sample in order to get a reliable number. This FanGraphs page says 820 balls in play are needed to get a good idea of a player's BABIP. Adding this into the Coors' advantage casts a lot of doubt on CarGo's BABIP abilities.

However, I didn't want to settle for just having doubts. I turned my attention to the Statcast data provided by BaseballSavant.com. While Statcast promises to provide all types of data, their hitter leaderboards currently incorporate Exit Velocity (the speed at which the ball leaves the bat) and Launch Angle (the vertical angle by which the ball leaves the bat). The data found here allowed me calculate the average BABIP for each possible combination of angle and velocity at which a batted ball has been struck.

Notably, a variable missing here is what Baseball Savant calls Batted Ball direction, which is the horizontal angle at which the ball leaves the bat. Baseball Savant doesn't include this in their leaderboards, so I was not able to include it in my analysis. However, we will be able to use proxy's for batted ball direction, and I'll get into that later.

First off, here's a neat heatmap I made of each angle and velocity combination, with the redder the area, the higher the BABIP, bluer the area, the lower the BABIP:

There's a lot to see here. Ground balls are those with a Launch Angle of less than 10, with line drives between 10 and 25, flyballs between 25 and 50, and pop-ups above 50. As you can see, hits between the angles of about 10 and 25 (the line drive angles) are very likely to be hits once they reach about 70 mph.

Balls hit in-between 60 and 70 mph and with an angle between about 25 and 40 are very likely to be Texas Leaguers, falling in-between infielders and outfielders. The large sea of blue to the right of the bloopers and above line drives is the "Donut hole", an area full of medium velocity fly balls that are extremely likely to be caught. To the right of the donut hole would be home runs, which are not included here as they are not balls in play.

Ground-balls do fairly well, as long as they have some decent combination of velocity and angle. The white region running diagonally and below the line drives illustrates that.

Let's break it down a little simpler. Let's look at the league BABIP just by angle:

There's a very obvious peak at 13 degrees, which has over a .800 BABIP. Ground-balls steadily become better as the angle increases, and once it hits 13 degrees it begins to decline.

Now, let's look at velocity:

For most this graph, more velocity is better. The exception from 70 to 90 mph is again the donut hole affect, though it's a bit suppressed as this is looking at all batted balls, not just flies.

On Baseball Savants' leaderboards, when you click on a players name it opens up a list of each batted ball along with some information, including Exit Velocity and Launch Angle. With that along with the above information, I was able to do some number crunching.

For instance, with the average BABIP of each angle, I was able to put together an expected BABIP (xBABIP) using only angles, in order to get an idea of how well a hitter's Launch Angle affected his BABIP, without consideration of Exit Velocity. I was also able to the do the opposite: figure out how much each hitter's batted ball's Exit Velocity, without consideration of Launch Angle, affected his BABIP. Finally, I found a total figure which took the expected BABIP of each batted ball's angle and velocity and used it to form a total xBABIP figure.

However, there was one issue: all of Baseball Savants' data leads to a league average BABIP of .324, quite a bit above .300, which is typically right around where league average BABIP sits. That's because Statcast has trouble picking up balls at extreme angles, which evidently are almost always outs. To fix this, I took the league average non-pitcher BABIP for 2016 (.302) and divided it by .324 to get a constant which I could apply to a hitter's xBABIP to properly scale it to league average BABIP. Here are the results for CarGo and Holliday:

Despite a Grand Canyon-sized gulf between the two hitters in BABIP, their xBABIP, at least so far as Exit Velocity and Launch angles are concerned, are strikingly similar. Holliday hits the hell out the ball, ranking third in average Exit Velocity so far this year, so his velocity-based number is about as high as you can get. Gonzalez is quite a bit lower but still posts very strong velocity numbers. Both hit at below average angles, but make it up by hitting the ball very hard. To get an idea where these two are hitting the ball most often, we can use a graphic from Baseball Savant which Joe used earlier in the week in his piece on Randal Grichuk. Props to Joe as I did not know about this feature until he used it. Here's Holliday:

Holliday has a lot of batted balls in between 15 and about 22 degrees, and another large chunk from -10 to 5. He's hitting the ball the hardest around those groups as well. The line drive group is going to have a rather large chance of being a hit as long as the hitter is making at least OK contact, and the groundball group has a decent chance of being hits when the ball is hit hard, which Holliday has been very adept at doing. Now CarGo:

Gonzalez has spikes around -5, 5, and 15. Two of those three have good chances of being hits, but the -5 group can only be hits when hit hard, and CarGo doesn't hit them as hard as Holliday. As we'll see in a second, he's hitting them into a shift fairly often as well, so even if he was hitting them as hard as Matt, we can't expect them to be hits as often. Like Holliday, he rarely pops-out. For velocity, Gonzalez is a bit more choppy in ranges that are good for BABIP, but in general his velocity holds up a bit better than Matt at ranges outside of those.

More things affect BABIP than velocity and angle of the batted ball though, and it would be wrong for that to be the endpoint of our analysis. Gonzalez has one point in his favor: he's faster, and thus more likely to leg out infield hits. Holliday has an advantage too though, as he's much less shift-able:

The "GB Spray Angle" is taken from Brooksbaseball.net. The foul lines are 90 degrees apart, and Brooks Baseball sets straight up the middle to 0 degrees, with the left-field line at 45 degrees and the right-field line at -45 degrees. The figure shown here is the average spray angle, so on average, Gonzales pulls his grounders more often than Holliday.

In case you didn't know (I didn't) the average speed score is 4.5, so both are slow, but Holliday is much closer to the bottom. That affect is dwarfed by how often Carlos Gonzalez's grounders are hit right into a shifted infield. He hits less grounders than Holliday does, so it's less important, but that still should be a significant drag on his BABIP that an angle and velocity-only system doesn't take into consideration.

Holliday and CarGo's xBABIP are neck to neck in terms of Exit Velocity and Launch Angle. When considering speed and shift data, it's easy to make the case that Holliday should have ended up with a higher BABIP than CarGo, despite that huge gap we currently have.

This doesn't necessarily mean Holliday will be better going forward, as Statcast began collecting data just a year ago and we don't know how repeatable these stats are quite yet. But when looking at what has actually happened, we know that Holliday's contact quality is way better than his BABIP implies.

Without that large gap, and the fact that his power numbers are certainly expected to decline when leaving Coors, CarGo suddenly looks like no upgrade at all when compared to Holliday. He looks like a fine hitter of course, but this is not a team in need of a corner-outfielder, and a team that is in need of one has a lot more reason to pony up the prospects it will take to acquire him. Carlos Gonzalez should be an easy pass for the Cardinals this month.