How Batted Ball Distance Ages

Jacoby Ellsbury might not recreate his 2011 season, but he is still bucking trends. (via Keith Allison).

Jacoby Ellsbury might not recreate his 2011 season, but he is still bucking trends. (via Keith Allison).

In 2011, Jacoby Ellsbury of the Boston Red Sox delivered a dynamic season at the plate, posting a 150 wRC+ fueled by a .230 isolated slugging figure and 16.7 percent of his fly balls going for homers. Never before had Ellsbury displayed that kind of power at the big league level, or even in the minor leagues over a significant number of at-bats. The question, of course, was whether this was a massive power breakout for the talented center fielder or if this was simply an outlier performance.

In 2012, Ellsbury was sidelined for more than half the season due to injury, posting pretty paltry numbers at the plate (84 wRC+, .099 ISO). He had a much better all-around year in 2013, but the power Ellsbury displayed was pedestrian compared with his 2011 campaign (.128 ISO, 6.6 percent HR/FB).

We can all think of similar scenarios in which a player, coming into his peak, puts up fantastic power numbers only to see those numbers come back down to earth as he ages. It got me thinking about the extent to which power — and the aging of power — could be studied by looking at batted ball distances, not just outcome statistics.

Methodology

Absent HITf/x data, I turned to my colleague Jeff Zimmerman and his PITCHf/x database. While not perfect, PITCHf/x is what we have available publicly. Also, it is possible to calculate batted ball distance and angle (i.e. field) based on coordinates provided through the GameDay data. The biggest caveat is that the coordinates mark where the ball was fielded, not necessarily where it landed. So, one can imagine a few data points in which the ball rolled to a deeper part of the park after landing. This, of course, would appear to be a deeper driven ball in the data.

I am using data from 2010 through 2013, so there are some limitations in terms of sample size (hence, the curves beginning with age-24 seasons). Pitchers were excluded from the sample, for obvious reasons. And I am looking only at batted balls that are coded as home runs or fly balls, and those that traveled at least 100 feet, and we are pushing the limits of the data. But, as I said before, until we have access to HITf/x we must make do with what we have.

In terms of breaking the batted balls into pull, center, and opposite field I used the following batted ball angles (numbers represent the degree of the angle relative to dead center field):

Batted Ball Angles
Area RHH LHH
Pull <= -15 >= 15
Center > -15 to < 15 >-15 to <15
Oppo >= 15 <= -15

For those familiar with our work on pitcher aging curves, the same method for calculating change by age cohort was used. In this instance, the harmonic mean of the first and second seasons’ number of batted balls was used for the weighting. (You can read more about the method here and here.)

Let’s dive in.

Aging Curves by Batted Ball Direction

AllFieldsAging

There are some slight differences in terms of how distance ages depending on what field the ball is hit to. Distance on balls hit to center appears to decline straight through a player’s career. Opposite field power sees a slight increase at age 27, and then begins a decline similar to center field power. Pull power, however, experiences an increase that is akin to what we’ve come to expect when looking at a player’s outcomes–peaking between ages 25-27, and then beginning a decline. Pull power is also more “choppy” than to the other two fields.

I would caution against reading too much into this for a few reasons. First, the data. We have to remember that we’ve got limitations with volume, quality, and of course the survivor bias that can’t be completely controlled for in using this method. Second, it could be that as players start to feel the effects of aging (slower pitch recognition, slower bats, etc.) they adjust their approach such that they are trying to hit the ball up the middle and the other way with more power than before.

Aging Curves by Handedness

If we split the curves based on batter handedness, we observe some slight, but interesting, differences.

First, if we look at pulled balls we see that right-handed hitters generally gain distance between ages 24-27 before experiencing a sharp decline. Left-handed hitters, in contrast, experience a slight increase between ages 25-26, but then start their decline as well. The key difference appears to be the pace of decline.

Between ages 31 and 35, right-handed hitters’ distance dramatically declines, while lefties seemingly stabilize into a more gradual decline at age 32. I’m not sure what’s behind the difference. Could it be that right-handed hitters  are able to get away with less power for longer? My initial guess would have been to say no (due to generally shorter porches in right field than left field across ball parks, speedy lefties may have an advantage productivity-wise without power that righties don’t have), but maybe I am missing something.

AllPulls

Let’s return to Ellsbury. In 2010, his average pulled home runs and fly balls traveled roughly 276 feet. During his breakout 2011, however, his average distance skyrocketed to 304 feet. Ellsbury added 28 feet on average to the balls that he pulled and elevated. In 2012, however, his pull distance declined to 280 and in 2013 he averaged 287.

Now, compared to 2011 he clearly suffered a decline. But relative to left-handed hitters over the same age span, that increase goes against the general aging curve. Lefties, from ages 26-29, generally lose about seven feet of distance when pulling the ball. So while Ellsbury isn’t hitting home runs at the same pace he was for that magical season, his physical power trend should be encouraging.

The patterns for opposite field distance are remarkably similar by handedness. Outside of the slight increase at age 26 for right-handed hitters, the two curves are almost identical. There is a slight bit of accelerated loss by left-handed hitters around age 31, but overall there isn’t a lot to tease out when it comes to going the other way.

AllOppo

Center field power is a little more interesting. Righties and lefties start off quite similar, but diverge after age 29. Left-handers see their distance decline at an accelerated rate relative to right-handed hitters.

AllCenter

Taking all three graphs together one gets the impression that, to stay in the majors and be productive, left-handed hitters might be consciously shifting their approach to driving the ball deep to their pull field. This makes sense when we think about the positions lefties (those who throw as well as bat left-handed) tend to play–first base and corner outfield. It’s hard to be a light hitter and stick around if you are playing one of those positions. Now, these curves aren’t looking  just at lefty throwers — you have switch hitters blended in to each curve–but the issue of position would explain some of the differences in the curves above.

Physical Changes in Power and Power Outcomes

We can also see parallels to pitcher aging. In our original study, Jeff and I found that the aging of pitching performance was tightly coupled to change in their fastball velocity. The relationship wasn’t perfect, however, and graphically it showed that pitchers can counteract the effects of a dying fastball. Essentially, performance aged more slowly than fastball velocity.

We see a similar relationship between batted ball distance — particularly, balls pulled by a hitter — and  performance from a power perspective. Below is an overall aging curve for pulled batted ball distance plotted against an ISO aging curve. The curves both cover 2010-213 and the same age buckets are represented to keep things consistent:

PulledvsISOAging

This graphic looks very similar to what we saw with pitchers. The raw physical skills were important (whether that be velocity or distance), but performance decline was not as sharp as physical skill decline.

It shouldn’t be surprising that a relationship exists, but I wanted to dig a little deeper and see the magnitude of the effect.

I matched up the distance of pulled batted balls and a player’s SLG, ISO, and HR/FB in the same year and ran some simple correlations. The results are pretty good and what we would expect:

Correlations between Distance and Outcomes (Same Years)

Correlation with Pull Distance
Statistic Correlation
SLG 0.52
ISO 0.55
HR/FB 0.60

At first blush, this aligns with the idea that distance can’t (and shouldn’t) explain everything about production, but it does explain a decent chunk. The numbers are also consistent with what Chad Young found about a year ago (although our methods are slightly different and he was not discriminating between all fly balls and pulled fly balls). But what about change in distance–how closely does that tie to outcomes?

Here, I matched up the year to year changes in distance with year to year changes in each of the three metrics above. The correlations are pretty good–not amazing, but a decent size and directionally what we would expect:

Change in Metrics and Change in Distance (same years)

Correlation with Change in Pull Distance
Change in… Correlation
SLG 0.29
ISO 0.34
HR/FB 0.40

Changes in pulled batted ball distance have the strongest correlation to changes in HR/FB. ISO came in with a .34 correlation, and SLG was not far behind at .29.

If we are going to gain anything explanatory or predictive out of the distance data it appears that pulled data is the only way to go. I also ran correlations for opposite field and center field distance, but the results were not  robust. In fact, the best correlation was for opposite field distance and SLG (.13).

In terms of predicting change, I took a similar approach but this time matched up the change in pulled distance from years 1-2 and correlated that with changes in each outcome metric from years 2-3:

Change in Distance (Time 1 to Time 2) & Change in Metrics (Time 2 to Time 3)

Correlation with Change in Pull Distance
Change in… Correlation
SLG 0.25
ISO 0.27
HR/FB 0.31

The results here are similar. The correlations aren’t incredibly strong, but they do tell a story that aligns with the aging patterns mapped out above. As distance decreases we should expect–in general–to see decreases in outcomes and power productivity.

Wrapping up

We’ve learned that batted ball distance declines as players ages and that seemingly impacts their production. Amazing!

All joking aside, I think the takeaway from this research is a better understanding of the rate of physical aging and the degree to which it impacts performance aging. To stick on major league rosters, hitters have to overcome their declining physical skills and find a way to still be productive. Much as with pitchers and velocity, they can’t simply rely on the raw power of their youth.

It also appears that left-handed hitters make more effort to focus on pulling the ball for power as they age, which is likely a function of the positions these hitters play and what is demanded of their offense. Right-handed hitters may have more flexibility in terms of their approach and may find it easier to remain on major league rosters even as their power declines.

In terms of player evaluation, the data suggest that tracking changes in batted ball distances–in particular, pulled batted balls–does add some explanatory power over current and near-future productivity. No doubt, front offices have already worked these data into their projection models. (And we are talking about even more telling data such as launch angles and speed off the bat.) For those of us without access to HITf/x, however, there still might be something to be gained by incorporating pull distance into our models.

As always, many thanks to Jeff Zimmerman for access to his data and his advice throughout this research. If you want to look up batted ball distances and angles by player, Jeff’s leaderboards can be found here.


Bill leads Predictive Modeling and Data Science consulting at Gallup. In his free time, he writes for The Hardball Times, speaks about baseball research and analytics, has consulted for a Major League Baseball team, and has appeared on MLB Network's Clubhouse Confidential as well as several MLB-produced documentaries. He is also the creator of the baseballr package for the R programming language. Along with Jeff Zimmerman, he won the 2013 SABR Analytics Research Award for Contemporary Analysis. Follow him on Twitter @BillPetti.
11 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
joe
10 years ago

You mention LHH might struggle to stay at a corner spot because of the hitting req’s. What about a guy who bats LH but fields RH (and thus has more options to play other positions). Hows his curves look like?

Mike
10 years ago

Wow, great stuff. The second chart on pull distance is intriguing. The RHH line seems to confirm what people generally think: that a player’s power increases and peaks at age 27. But this is not true for LHH hitters. But RHH’s pull power drops off consistently and sharply after age 27, while LHH are hitting the ball still relatedly well to pull. Is it something about bat speed for RHHs? Can they not turn on pitches as much as they used to? Do they learn to go the other way more? Is there something about those sweet LHH swings that allows them to keep getting to those inside fastballs and hitting them well into their 30s? Really interesting stuff. There are a number of follow up analyses I’d love to see, but I can imagine this impacting the market for free agent RHHs known for pulling the ball that are entering their early 30s. they better have demonstrated a skill for hitting the ball well the other way if they expect to get paid, bc that pull-power drop off is significant and progressive.

Peter Jensen
10 years ago

Are you defining fly balls using Retrosheet (MLB) definitions or BIS which includes infield flies?

but maybe I am missing something.

What you are missing is that left handed hitters are predominately 1B, DHs and outfielders unless they are switch hitters. These positions almost require power hitting ability because there is little chance to add value through defense. Conversely, catchers, 2B, 3B, and SSs take more than twice as many PAs as right handers than they do as left handers. Those players tend to be smaller, lighter, and have less power hitting ability. The ones that age best are able to create value through defense and higher average hitting rather than power hitting. Hence the selection bias evident in your data.

Roger
10 years ago

You focus on average batted ball distance. Is there any change in variance? Could players selling out for home runs as they age be seeing a wider variance in their batted ball distance?

I do fear that the overall trend towards more dominant pitching / less offense over the past few years could be outweighing any natural positive aspects of the hitter’s aging curve. All of the data have in common that the latter years represent a less hitter-friendly playing environment.

bob
10 years ago

The numbers don’t add up for me. Because, looking at the leaderboard, the range in fly ball distance for 2013 for all but the most extreme individuals is roughly from 300 feet down to 260 feet for a difference of 40 feet. In other words, David Ortiz is 302 feet and Alcides Escobar is 258 feet and only a handful are above or below those two. And you are claiming that 40 feet is also the average decline for a player aging from 24 to 37, if I understand this correctly? So the average player declines from hitting like Ortiz to hitting like Escobar over his career?

In other words, when the range from best to worst in that statistic during a season is hardly more than 40 feet for most players, I don’t see how an average player can decline from one of the best to one of the worst during a career. There can’t be a significant number of players who decline much more than 40 feet because most of them won’t be in baseball any longer after falling that far. I would therefore expect survivor bias alone to make the average career decline much less than 40 feet.

channelclemente
10 years ago

I wonder if one drew on the time of flight data, you could extend this analysis to look at a surrogate for backspin’s effect on distance.

Peter Jensen
10 years ago

When you aggregate MLB Gameday fly ball distance from a large number of hit balls by many batters the resulting averages are relatively accurate. When you aggregate balls hit by a single batter the average is much less accurate because of poor park drawings used by MLB to record the data and because of actual differences between parks (think of balls hit off the left field wall in Fenway) and also because of differences in the types of fly balls hit by different batters. When you look at Gameday distance for individual hit balls the recording error is too great to do backspin analysis or really any analysis at all.

Dirck
10 years ago

The very steep drop between ages 27 and 28 looks like an anomaly . I believe you say this was numbers from 2010-2013. Is it possible that there were several 27 year old big power hitters who were injured and hugely underperformed their earlier numbers during their age 28 season
and thus significantly skewed the numbers .Separating the figures into so many age groups makes each age sample size much smaller than you would think at first glance .

Brendan
10 years ago

You and the rest of the writing staff are killing it with this research lately. Keep up the good work. I’ve used BBIP distance for a while and always wondered about the correlation, but was too lazy to do what you just did. This is beautiful.