The essence of hitting, with all its nuances and intricacies, can be summarized succinctly as “barrel to ball.” From little league to the major leagues, if a hitter solidly squares up a pitch, good things will generally happen. If they don’t barrel it up…not so much. Thanks to Baseball Savant’s bat-tracking data, this maxim can now be verified precisely at the MLB level…

What is a squared-up swing? Savant defines it as a batted ball that reaches 80% or more of the estimated maximum possible exit velocity, a function of both the swing speed and pitch speed, for a given swing. For example, on a swing with a maximum possible exit velocity of 100 miles per hour, an 85 mph batted ball would be counted as “squared up,” while a 65 mph ball would not. In practice, most full swings have a maximum EV of ~100-110 mph and therefore hard-hit balls (with a 95+ mph EV) are generally classified as squared-up. (Note: This piece will refer to the percentage of maximum exit velocity on a given swing [the play level] as “barreled-up rate” or “% ball is barreled up” to distinguish this metric from “squared-up rate,” which has been used to refer to the percentage of a batter’s swings that are above the Savant squared-up threshold [the player level]).

Subscribe now

Share Down on the Farm

Using individual pitch data, we can visualize the distribution of barreled-up rate for balls in play so far in 2024…

At first glance, it may be surprising that the vast majority of balls in play are barreled-up at over 70%. Keep in mind that this metric refers to the percentage of maximum EV for that swing, NOT some measurement of the percentage of the bat’s sweet spot that the ball connected with (though that is surely related to barreled-up rate causally). This fact, combined with the focus exclusively on balls in play, without whiffs and foul balls, explains the elevated percentages.

Unsurprisingly, barreled-up rate has a strong relationship with common quality of contact metrics, including exit velocity and expected wOBA…

Exit velocity increases nearly linearly with barreled-up rate. However, there is a small dip in EV for balls barreled up ~95% or more. This is because bat speed is typically lower at this high barreled-up rate level (more on this in a moment). The relationship between barreled-up rate and xwOBA is less straightforward. Expected wOBA remains low for barreled-up percentages below 80-85% (hence Savant’s 80% squared-up threshold) before rising rapidly through ~95%, with a similar dip to the EV plot above 95% barreled-up. This non-linear relationship is understandably reminiscent of the link between exit velocity and wOBA. In simple terms, barreling the ball squarely is key, while differing levels of mishits have largely similar expected results.

Barreling the ball is key to a given batted ball’s success, but the swing’s bat speed at contact is certainly another important variable. The optimal combination of a high barreled-up rate and high bat speed is quite lucrative, but also relatively rare…

These treasured batted balls are codified by Savant as “Blasts,” and are largely the territory of baseball’s premier sluggers, including Aaron Judge and Shohei Ohtani. Their rarity is due in part to baseball’s version of “no free lunch.” Barreled-up rate is typically lower at the highest swing speeds…

Striking a tiny projectile thrown at blindingly fast speeds in the center of a thin wooden stake is quite difficult. Doing so at bat speeds unnatural for a given batter is especially challenging, particularly at swing speeds significantly higher than the player’s norm. This fact seems to be the driving force of the relationship between raw bat speed and barreled-up% and generally holds up even when controlling for other factors (such as pitch type and location, count etc…). Players tend to maximize barreled-up rate on bat speeds and launch angles near their overall averages…

The combined effect of these two variables on barreled-up rate can be modeled with a generalized additive model…

Batted balls are most often barreled-up at bat speeds and launch angles most similar to the player’s norms. Barreled-up rate drops farther from the midpoint, and does so especially precipitously at high bat speeds and low launch angles. As a result of all of this, players with high typical bat speeds and optimized swing planes have a leg-up in taking impactful swings without sacrificing their ability to barrel up the ball.

Now that the importance of barreling up the baseball is clear, a logical follow-up question is: “How best can we determine which players are good at squaring up their batted balls?” As previously mentioned, Baseball Savant uses the logical 80% barreled-up threshold to classify a swing as squared-up. This is far from the only usable threshold though. Perhaps 60% or 90% are thresholds more indicative of a batter’s barreling skill. A reasonable method to determine a threshold’s utility is to evaluate how stable the general performance of batters is at various sample sizes. This can be done using split-half correlation…

It seems like most high squared-up thresholds (80% and above) all tend to have a reasonable amount of stability between samples, with the 95% threshold pulling away from others in larger samples. However, the most stable measure of barreled-up performance does not seem to be a threshold metric at all and is instead average barreled-up rate (for those interested, the ballast/regression amount seems to be ~30 batted balls). Perhaps the more granular mean measurement simply conveys more information than a binary threshold can. Unsurprisingly, bat-control wizard Luis Arraez tops the list for most of these metrics, while Daulton Varsho and Jo Adell are some of the notable laggards.

High squared-up thresholds (following average barreled-up rate) are the most stable indicators of player skill, but this does not mean that lower thresholds are useless. Each threshold, by including or excluding various batted balls, has a performance “fingerprint.” To illustrate what I mean, it’s important to bring another variable into the fold: Launch angle. As Ben Clemens has astutely pointed out at Fangraphs, launch angle is an important lens through which to view squared-up rate. Barreling up the ball is not nearly as important on grounders as it is on high fly balls. Taking this a step further, barreling the ball at 95% or more is particularly important on flies. The 80% threshold is not nearly as informative…

Red Dotted Line = Overall Average xwOBA on Contact

On the other hand, for low liners in the ~8-12 degree range, the average outcome for balls over the 80% threshold is nearly indistinguishable from 95%.

Given the interconnected nature of barreled-up rate and launch angle, a single metric combining both variables may be useful. In a sense, this metric may be thought of quantifying the value of a batter’s contact independent of their bat speed, though bat speed may still have secondary impacts on both metrics. Using a gradient boosting model (as I did for xwOBA on contact) or other complex model is natural, but in the spirit of simplicity and the earlier generalized additive model, let’s use a GAM instead. Comparing a grid of the predicted and actual results, it seems that the GAM does a nice job of capturing the effects of launch angle and barreled-up rate on wOBA…

This metric, tentatively named the (mouthful) “Bat Speed Independent xwOBA on Contact,” seems to be a relatively stable player stat, with a regression amount of ~45 average batted balls. It also correlates relatively well (r = 0.43) on the player level with typical xwOBA on contact, unlike squared-up rate per swing. Once again topping the leaderboards is the Padres’ Luis Arraez…

Only Batted Balls with Swing Speed. xwOBAcon may Differ from Savant Numbers

Arraez rates highly not only due to his exceptional ability to barrel-up the baseball, but also because he does so at optimal angles, as evidenced by his excellent Sweet Spot%. The same is true of many other leaders. Michael Busch is a particularly interesting case. His squared-up rate on contact is approximately average, however much of his squared-up contact occurs on fly balls where their value is maximized. Returning to Arraez, he also leads the way in a far more ignoble category. The negative difference between Arraez’s Actual and Bat Speed Independent xwOBAcon is the largest of qualified batters. Arraez has sublime bat control, but his lack of bat speed robs him of many extra-base hits. In sharp contrast to Arraez stands the exceptional Aaron Judge. Judge pairs two remarkable skills. He is among the leaders in Bat Speed Independent xwOBAcon and piles on top of that plus bat speed, which when combined produces unparalleled contact quality.

On the other end of the spectrum, here are the metric’s laggards…

This list is littered with talented players with plus bat speed. Baez, Carroll and Torkelson, to select just a few, have xwOBAcon numbers that are shockingly low given their raw power. At the end of the day, elite bat speed can act as a useful jolt for well-struck fly balls, but without the raw material to act on, its utility is limited. With the notable exception of Giancarlo Stanton (who ranks near the bottom in BSI xwOBAcon), both bat speed and ideal contact are key to posting elite contact quality. A tentative BSI xwOBAcon leaderboard can be found here.

To review:

  • The vast majority of balls in play reach >70% of their maximum possible exit velocity (barreled-up rate).

  • Barreled-up rate has a mostly linear relationship with exit velocity.

  • xwOBA on contact rises dramatically on batted balls past 80-85% barreled-up.

  • Batters typically achieve their highest barreled-up rates at bat speeds and launch angles near their averages.

  • Average barreled-up% seems to be the most stable indicator of player barreling ability, followed by high squared-up thresholds.

  • It is especially important to have a barreled-up rate near or above 95% on fly balls and is much less important on low line-drives or grounders.

  • Bat-Speed Independent xwOBA on Contact seems to be an effective measurement of a batter’s ability to produce ideal contact.

Statcast bat-tracking data makes it possible to isolate the components of batter performance more potently than ever. In particular, Bat-Speed Independent xwOBAcon is a metric in its infancy. Further refinement and research on its merits and flaws are needed. Feel free to contact me (@matan_writes) with any questions, thoughts, or suggestions.