Modeling the Effect of Deadening the Baseball

Much has been made of the “juiced ball era” which we currently inhabit. Decreased drag on the ball along with an increase in-ball bounciness means that fly balls are carrying further, rewarding hitters with more home runs than ever before. This change has coincided with increases in strikeout rates which can be partially explained by pitchers throwing harder, but also may be due to more hitters selling out for a home run. There are now fewer balls in play than ever before, and many fans no longer enjoy this Three True Outcomes style of baseball.

Deadening the ball is a proposed solution to ballooning home run rates. Introducing a deadened ball along with measures to limit the dominance of pitchers (such as shrinking the strike zone) could increase the number of balls in play, improving the aesthetic value of baseball for many viewers as discussed on this site in a recent article. But what would baseball with a deadened ball actually look like? How much would the ball have to be deadened to return home run rates to those seen in past years? Would deadening the ball disincentivize strikeouts more strongly than the juiced ball? Which hitters would be the biggest winners and losers in a season with a deadened ball?

I aim to investigate all these questions in this article, so without further ado, let’s dive right in.

Modelling a Deadened Ball

We can’t replay a season with a deadened ball, so I’ll have to get a little creative to predict the effect of deadening the ball. To do this I’ll be modelling outcomes based on the velocity and launch angle of a batted ball. Baseball Savant is kind enough to provide these details for almost all batted balls. In addition they provide expected statistics such as batting average (xBA) and wOBA (xwOBA). With these I can predict the results of batted balls hit with different properties. xBA tells us the chance of a ball hit with a given launch angle and exit velocity falling for a hit, which is done by averaging the results from balls hit with similar trajectories.

Unfortunately, xBA and xwOBA are not predictive models; I can’t provide new batted balls and have Baseball Savant provide xBA and xwOBA values for them. Therefore I created my own models to predict xBA and xwOBA using a simple machine learning technique called XGBoost, which I have used in a previous article here. The figure below shows that my model xwOBA looks very similar to the original xwOBA, so we can use this to predict the expected results when I simulate using deadened balls. The xwOBA errors are small and not biased in a particular direction. I extended these models to also provide xSLG and xHR, in addition to xBA and xwOBA, for a more complete model of how deadening the ball would affect batters’ performances.

With these tools, it is possible to model the effect of the deadened ball.

A Less Bouncy Baseball

Making the baseball less bouncy would reduce its coefficient of restitution. The collision between bat and ball would not transmit as much energy to the ball, which reduces exit velocities. The exact reduction of exit velocity would depend on a complex mix of bat speed, ball speed, and quality of contact. In this case, I’ll assume that a reduction in ball bounciness applies a constant fractional reduction in exit velocity to all batted balls. For example, making the ball 5% less bouncy turns a 100-mph batted ball into a 95-mph batted ball and turns an exit velocity of 50 mph into 47.5 mph. With the models for xBA, xwOBA, xSLG, and xHR, it’s possible to find the projected outcomes using a deadened ball.

Let’s use the 2019 season as our test case. I took all the batted ball events in 2019, applied an exit velocity reduction to simulate using a deadened ball, ran the deadened balls through the predictive models, and collected the results. About 10% of batted balls were missing Statcast data and therefore had to be omitted from the results. Here’s how the league’s simulated slash line varies with ball bounce reduction, along with the change in home run rate:

If 2019’s Ball Were Less Bouncy
Bounciness Reduction BA Change OBP Change SLG Change Home Run Rate Change
1% -.007 -.006 -.026 -15%
2% -.013 -.012 -.051 -28%
3% -.020 -.017 -.075 -41%
4% -.025 -.022 -.096 -52%
5% -.030 -.027 -.115 -62%
6% -.035 -.031 -.133 -71%
7% -.039 -.034 -.147 -78%
8% -.042 -.037 -.160 -83%
9% -.044 -.039 -.171 -88%
10% -.046 -.041 -.179 -91%

Home runs and slugging percentage are most affected by reducing ball bounciness. Making the ball 4% less bouncy would halve the number of home runs. Reducing the ball bounciness by 7% would cause slugging percentage to fall below on-base percentage, which has never happened before in MLB history. It is clear that minuscule changes to the ball can have very large effects and that making the ball a bit more or less bouncy would change the game considerably.

Deadening the ball has been theorized to reduce strikeout rates by forcing hitters to rely on a more contact-oriented approach. This can be seen when comparing the correlation between hitters’ strikeout rates and their xwOBA as the ball gets deadened.

The current ball gives a negative correlation between xwOBA and K% of around -0.13. As the ball becomes deader, this negative correlation becomes much stronger, and if the maximal value of a batted ball becomes lower then it is important to get as many in play as possible. However, extreme changes are needed to shift this correlation, so it looks like deadening the ball by a realistic amount will have little impact on strikeout rates.

MLB has announced that they are deadening the baseball to reduce home run rates by around 5%. In my models, this corresponds to a bounciness reduction of 0.3%, a very small change. This would reduce batting average by two points and slugging percentage by nine points. Using my model of the 2019 season, we can investigate which hitters would have been affected the most by this change.

If 2019’s Ball Were Less Bouncy
Player xwOBA change xHR change
Mitch Garver -.0150 -1.0
Cavan Biggio -0.011 -1.0
Aaron Judge -0.010 -0.8
Nick Castellanos -0.010 -1.9
Rougned Odor -0.010 -1.7

My model predicts that the change MLB is proposing is unlikely to shift batters’ wOBA by more than 10 points and will take at most one or two home runs from sluggers over the course of a season, almost imperceptible.

A New Deadball Era?

As a bit of fun, let’s imagine what would happen if the scientists at Rawlings got their calculations drastically wrong and removed far too much juice from the baseball recipe. The resulting lifeless creation is 10% less bouncy than its predecessors. Refusing to admit that they made a mistake, MLB insists that the new ball is used for the entire season.

Using the 2019 season as a source for batted balls, Jorge Soler would lead the league in home runs with a whopping 10, matching Home Run Baker’s mark in 1912. Luis Arraez takes home the batting title with a sizzling .280 average and is second in wOBA behind the evergreen Mike Trout.

The changes that will be made for the 2021 season won’t bring us back to baseball in the early 1910s. There will be fewer home runs but no other changes to provide compensation for hitters, while strikeout rates are likely to continue rising. I’m glad to see the league making proactive changes to make the game a better product, but as we can see, making changes to the baseball needs to be very finely tuned.

newest oldest most voted
D.K. Willardson

Interesting research. On the strikeout side, I think the deader ball 2018 (actually a drag issue) may give a decent indication that Ks may not be effected at all w/ the new ball. Done quite a bit on Attack Angle vs. K% and that seems to be the main connection between xWOBA and K rate. So unless you get a reversal in league-wide AA, K rates likely are going to continue the existing trend higher. There is also a connection between AA and EV- independent contact quality – so based on the data, it would seem unlikely that hitters are… Read more »

Mario Mendoza
Mario Mendoza

Cool article. Now a week into the season, it hardly seems the ball is even 0.3% less bouncy.