Exit Velocity Carryover Effect

Of all the statistical advances made in the recent past, exit velocity seems to get the most attention. Broadcasts that still shy away from discussions of WAR or wRC+ or UZR are readily using exit velocity on batted balls. Part of that could be the novelty of it, and part of it is just a fascination with how hard and how far a ball is capable of being hit. Part of it could also be a sort of familiarity. Home-run distance has long been included in broadcasts, as has been a pitcher’s velocity. Exit velocity is an easy expansion of those numbers.

That said, exit velocity isn’t just a novelty. Despite issues with the data and the importance of launch angle and batted-ball data as a means to providing context, a player’s average exit velocity can tell us a decent bit of information about a player. With another half of data available (thanks to our own Jeff Zimmerman for his assistance gathering data), we can attempt to determine whether exit velocity from last season carried over to this season.

When I looked over the winter, the correlation on an individual player level between wOBA and exit velocity was relatively strong (r=.61) over the course of 2015 for players who had played a majority of that year. That number is not as strong so far this year (r=.50), but we are also dealing with a larger (237) universe of players with a lower level (200) of plate appearances over the first half of this season. It will be interesting to see if the correlation climbs a little higher as the season continues.

Last year, there was a solid relationship between first-half and second-half exit velocity. To determine how much of last year’s numbers carried over to this season, I compared the 116 players who recorded at least 200 plate appearances in each of the last three half-seasons. First-half exit velocity from 2015 correlated well with first-half exit velocity for 2016 (r^2=.52), but not as well as second half of 2015 with first half of 2016 (r^2=.57). The strongest relationship between the periods was between the entire 2015 season and the first half of 2016 (r^2=.62).

Exit Velocity Carryover from 2015 to 2016

If you want the best bet for what a player will do this year, looking at a full year of data is the way to go based on the information we have, but we don’t know if that is uniform for all players. What about the players who experienced changes from the first half of 2015 to the second half of 2015? Did those changes carry over? Yes and no.

  • For the 33 players who had large increases in the second half last year (at least 1.5 mph increase), the second-half exit velocity had a slightly higher correlation than 2015 total (r^2=.51 compared to .47). A good second half of exit velocity might be a harbinger of continued higher numbers.
  • For the 23 players who produced a decrease of at least 0.5 mph, the decrease seemed to have less bearing, as there was a smaller correlation (r^2=.38) for the second half of 2015 to the first half of 2016 compared to 2015 as a whole to the first half of 2016 (r^2=.44). A dropoff in the second half in terms of exit velocity is less important than the full year of numbers, it would seem.
  • For the players who had relatively consistent halves in 2015, those numbers have carried forward to 2016 (r^2=.74).

When I looked at the numbers over the winter, I was hoping to find some sort of application for the data I found. Everything else is fun to figure out (depending on your definition of fun), but to find something with utility would be most interesting. I looked at the population of players last season whose wOBA seemed to underperform or overperform their exit velocity — i.e., players who’d recorded above-average exit velocity but below-average numbers, and vice versa. I found that those players who underperformed their exit velocity in the first half saw their offensive numbers rise in the second half. Similarly, players whose offensive numbers seemed to overperform their exit velocity tended to have weaker number in the second half. Taking a look at those players for the first half of the season is probably worth a post on its own, so we’ll hold off on that for now.

What we will do now is take a look at players who over- or underperformed their exit velocity over the entire course of last season and determine if it had any carryover effect to this season. For a refesher on how I determined performance, here is what I said in February:

I created IQ-type scores for exit velocity and wOBA from the first half of last season based on the averages of the 130 players in the sample. In each case, I assigned a figure of 100 to the sample’s average and then, for each standard deviation (SD) up or down, added or subtracted 15 points.

Once the IQ scores for both stats were calculated, I subtracted the IQ score for exit velocity from the IQ score for wOBA to find the players with the biggest disparities.

In this case, I used the 116-player sample of players from above and looked at all of last season. We will tackle the underperformers first. It would seem the players below should have done better last season based on exit velocity. The chart below shows the wOBA IQ minus the exit velocity IQ for 2015, 2015 wOBA, projections for 2016 wOBA, 2016 wOBA, and the difference between the 2016 wOBA and the preseason projections.

2015 Exit Velocity Underperformers
2015 wOBA-EV IQ 2015 wOBA 2016 proj 2016 wOBA Difference from Proj
Wilson Ramos -41.7 .265 .295 .390 .095
Ian Desmond -27.9 .294 .309 .382 .073
Michael Taylor -23.1 .274 .287 .279 -.008
Kurt Suzuki -20.4 .269 .286 .334 .048
Gregory Polanco -19.2 .304 .311 .361 .050
Joc Pederson -18.6 .335 .329 .341 .012
Melky Cabrera -17.8 .307 .321 .341 .020
Jay Bruce -17.8 .309 .321 .354 .033
Brett Lawrie -16.4 .306 .321 .311 -.010
Albert Pujols -15.1 .333 .335 .318 -.017
AVERAGE -21.8 .300 .312 .341 .030

On average, these players have beaten their projections by 30 points, with seven of 10 up 10 points or more and six of ten up 20 points or more. Most of the players listed had down seasons last year and so projections had them bouncing back some, but the results so far this year have exceeded those projections. Wilson Ramos has benefited from better eyesight, but it is also possible his bad luck last season turned around a bit this year. While the current numbers put up by Ramos or Desmond are unlikely to continue quite like this, they still have a ways to drop and still be well ahead of this year’s projections. The exit velocity for the players above is roughly the same as last year, with only Bruce (-1.4) and Taylor (-1.8) changing by more than 0.6 mph from last year.

The overperfomers from last season tell a bit of a different story. Using the same methods as above, the players below had the biggest positive differential between their wOBA IQ and exit velocity IQ. THe chart includes the same information as the chart above, except the change in exit velocity from last year has been added.

2015 Exit Velocity Overperformers
wOBA-EV IQ 2015 wOBA 2016 Proj 2016 wOBA Diff from Proj EV CHange from 2015 (MPH)
Bryce Harper 40.7 .461 .430 .369 -.061 -1.3
Billy Burns 37.1 .317 .294 .249 -.045 -0.3
Joey Votto 34.4 .427 .392 .357 -.035 0.5
Jose Altuve 27.4 .347 .335 .400 .065 2.6
Ender Inciarte 25.2 .325 .305 .264 -.041 -0.7
Ch. Blackmon 22.9 .345 .335 .371 .036 1.8
Anthony Rizzo 22.8 .384 .373 .419 .046 1.2
Ian Kinsler 21.2 .335 .320 .358 .038 1.1
Matt Carpenter 18.4 .372 .347 .414 .067 2.3
Kevin Pillar 17.2 .310 .308 .301 -.007 3.5
AVERAGE 26.7 .362 .344 .350 .006 1.1

The players who seemed to overperform last year didn’t have a dropoff in performance, putting up roughly the numbers expected of them on average. What is interesting to note is that the players who have performed better also saw a decent increase in their exit velocity this year. Only Kevin Pillar, who even with the increase doesn’t hit the ball exceptionally hard, posted an increase of more than 1.0 mph and failed to see an increase over both last season as well as the projections. The players who did not up their exit velocity — like Bryce Harper, Joey Votto, and Ender Inciarte — saw production declines from last season.

Whether the underperformers breaking out the next season or the overperformers increasing their exit velocity is something that holds up will take further study down the line once we have more information. Simply using average exit velocity does have its problems. Launch angle and batted-ball type are very important in determining what a batted ball will do, and we are getting more information that could help down the line. However, producing exit velocity as a hitter certainly looks to be a repeatable skill and one that holds up from year to year based on the early data we have — and could have applications for future use, especially when it comes to identifying outliers and potential future performance.





Craig Edwards can be found on twitter @craigjedwards.

3 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
AA1989
7 years ago

You should take the average of the absolute difference from the projection. Taking the average of positive and negative differences make your projections look more accurate than they actually are. Otherwise, cool stuff.