Which Player Has The Best Swing?
Using bat speed and swing path metrics to measure swing quality.
The landscape of baseball analysis, on both the public side and within Major League organizations, has changed dramatically over the past decade, with technological advancements such as the implementation of Statcast and Hawk-Eye allowing for a greater emphasis to be placed on how and why an event within a game occurred, as opposed to simply recording what transpired. This evolution has been especially prevalent in regard to evaluating pitching, with pitch quality models, such as Stuff+, becoming prevalent tools in evaluating a pitcher’s underlying “true talent”.
While there have been lots of developments in implementing these metrics to improve pitching development, there was little development in creating predictive metrics for offensive players until Statcast unveiled their first batch of bat tracking metrics, bat speed and swing length, in May 2024, later releasing bat path data, such as swing path tilt and attack angle, in May 2025. Bat speed has long been a trait that scouts have looked for when evaluating hitters, and the rollout of this data has allowed the public to understand the predictive power of simply evaluating a player’s bat speed, confirming scouts’ longstanding intuition that swinging the bat fast is a good portender of future success. Swinging the bat fast leads to hitting the ball hard, which leads to better outcomes on balls hit into play.
As mentioned earlier, Stuff+ has been a powerful tool that can be used to evaluate pitchers solely off their velocity, movement, and spin characteristics, and I have long hypothesized that a “Swing+” metric, evaluating a hitter purely off their bat speed and swing path, could be an equally powerful tool for evaluating offensive players. With bat speed and swing path data now publicly available across multiple seasons, dating back to the 2023 All-Star Break, I have been constructing the first version of my swing quality model, aSwing+, which I am excited to release with the publication of this article. aSwing+ is a predictive metric that evaluates a hitter’s raw power potential using bat tracking data, incorporating bat speed, swing path, and contact point variables to estimate the probability that each swing will result in ideal contact. In this article, I will discuss the methodology behind constructing the model, discuss which features are important when evaluating swings, evaluate which players possess the best swings in Major League Baseball, and discuss the limitations and promise of utilizing aSwing+ moving forward.
Overview:
Before proceeding with a discussion regarding the model’s construction, let’s begin with a brief primer on Statcast’s bat tracking. Bat speed is measured as the velocity of the “sweet spot” of the bat at a swing’s point of contact (or estimated point of contact), while swing length measures the length of a swing from its starting point to the observed or estimated point of contact. Swing path tilt measures the angle of the player’s bat path over the 40 milliseconds prior to contact, with a higher angle reflecting a “steeper” swing and a lower angle reflecting a “flatter” swing. The swing path tilt of a given swing can be influenced by the pitch’s location, as pitches up in the zone are likely to generate flatter swings while pitches lower in the zone are likely to generate steeper swings. Attack angle measures the vertical direction of a player’s bat at the moment of contact (or estimated point of contact), and is heavily influenced by intercept point (the estimated point of contact relative to a hitter’s center of mass), as the bat travels more upward throughout a player’s swing. Attack direction measures the horizontal direction of a player’s bat at the moment of contact (or estimated point of contact) and, similar to attack angle, is also influenced by a hitter’s timing.
aSwing+ takes a “kitchen sink” approach to analyzing each player’s swing quality, using each of these bat tracking metrics (bat speed, swing path tilt, attack angle, attack direction, contact point, swing length), along with interaction variables and the given location, vertical approach angle, and horizontal approach angle of each pitch to predict the probability that each swing will result in an ideal contact outcome for the hitter. The model, a CatBoost classifier model, is trained on all batted balls that contain bat tracking data from the 2023 and 2024 seasons, with predictions made on all swings with the requisite data present from the 2023, 2024, and 2025 seasons. For more information regarding the methodology of the model’s construction, as well as a detailed summary of how the model is created, please refer to my article Methodology: aSwing+.
The goal of the aSwing+ model is to possess better descriptive and predictive performance at evaluating a player’s offensive performance than simply looking at a player’s bat speed. Simply analyzing a player’s bat speed can provide an analyst with a lot of information regarding a player’s offensive potential in a small sample size, and it was my intention to improve upon this performance by evaluating bat speed in tandem with swing path metrics to obtain a better understanding of how a player generates their offensive production. To identify whether or not the model obtains this objective, I analyzed how well aSwing+ describes notable offensive metrics compared to a player’s average bat speed during the 2025 Major League Baseball season (through August 31st).
As shown by the table above, aSwing+ exhibits better performance at describing a player’s wOBA, wOBAcon, xwOBAcon, and ISO than simply using a player’s average bat speed. In my opinion, this result makes intuitive sense, as bat speed simply is a measure of a player’s exit velocity potential, while taking into consideration swing path metrics allows for a player’s ability to hit the ball in the air to be taken into consideration as well. While hitting the ball hard is good at all launch angles, hitting the ball hard and in the air is the best batted ball outcome, and I believe that aSwing+ is the best metric available that is able to measure a player’s ability to achieve this feat by simply taking into account their swing characteristics.
While displaying better descriptive power than average bat speed is a positive attribute for the aSwing+ model, bat tracking metrics are primarily used as predictive metrics (to assess a player’s “true” talent level and predict their level of future offensive production), and it is imperative that aSwing+ also performs well in predicting year-to-year offensive production. As shown by the table above, Y1 aSwing+ exhibits better performance at predicting a player’s Y2 wOBA, wOBAcon, xwOBAcon, and ISO than simply using a player’s Y1 average bat speed. aSwing+ is also very “sticky” year-to-year, although average bat speed is able to predict itself slightly better. While average bat speed still provides very useful information and can be a simple way to evaluate a player’s offensive potential, aSwing+ displays superior descriptive and predictive performance at measuring a player’s raw power capabilities.
2025 Leaderboards:
The table above depicts the players with the top 10 highest aSwing+ grades during the 2025 Major League Baseball season, minimum 200 swings. aSwing+ is scaled on a 100 mean / 10 standard deviation scale, relative to the average player’s swing quality. For example, a player with a 120 aSwing+ possesses a swing that is two standard deviations better than the average Major League player (in terms of scouting grades, this would be a 70 raw power). Triston Casas displayed the most powerful swing in Major League Baseball this season, with a 132 aSwing+, and the leaderboard is filled with some of the best power hitters in the entire league. Nick Kurtz and James Wood both appear among the top 10 swings in the league, providing optimism that they will remain power-hitting stalwarts for the foreseeable future, while Jake Bauers makes perhaps a surprising appearance on the leaderboard with the 3rd-most powerful swing in the league at 131 aSwing+.
The table above depicts the players with the 10 lowest aSwing+ grades during the 2025 Major League Baseball season, minimum 200 swings. This leaderboard contains players with some of the lowest bat speeds in the game, such as Chandler Simpson, Nick Allen, and Jake Mangum; however, the leaderboard also contains bat-to-ball savants such as Luis Arraez and Jacob Wilson. This underscores the notion that aSwing+ is primarily a measure of a player’s raw power ability, and players with excellent bat-to-ball skills can still generate above-average offensive production despite possessing a low aSwing+.
The table above depicts 5 notable players who experienced a significant increase in their aSwing+ from 2024 to 2025, with Brice Turang improving his swing quality from an 83 aSwing+ in 2024 to a 99 aSwing+ in 2025. This increase for Turang is largely driven by an increase in his average bat speed from 66.2 MPH to 70.6 MPH, and an increase in his average attack angle from 1 degree to 7 degrees, and this improvement in swing quality has allowed him to generate quality contact in the air more often (increasing his barrel rate from 2.4% in 2024 to 7.9% in 2025) and produce a career-high .148 ISO this season. While Anthony Volpe has had an inconsistent season at the plate, his 3.3 MPH increase in average bat speed also resulted in a significant aSwing+ increase (from 93 in 2024 to 108 in 2025), which has allowed him to produce more power this season. George Springer’s resurgent 2025 season can be partially attributed to his improved aSwing+, while Josh Bell and Ryan McMahon have improved their barrel rates while underperforming their expected offensive production.
Interpretations:
While I am certainly elated with the performance of the aSwing+ model, the real value in the model comes not from simply utilizing the outputs of the model, but rather from interpreting how the inputs interact with each other to obtain a given result. Machine learning models, such as CatBoost, are “black box” models, and it is difficult to fully understand how the model arrives at its conclusions; however, we can create some visualizations displaying the relationship between aSwing+ and its various features to better understand the factors that contribute to a valuable swing.
Given its strong correlations with various offensive metrics displayed earlier, it should be no surprise that average bat speed is a very important factor in determining a player’s aSwing+. Similar to how throwing a four-seam fastball 100+ MPH will cause the pitch to grade out well in Stuff+ regardless of its shape, players with high average bat speeds will always grade out favorably in the model. Much has been written about the importance of bat speed on this blog and in other publications over the past couple of seasons, and its heavy importance in determining a player’s aSwing+ is another reason why training for bat speed should be an important piece of each hitter’s training.
The scatter plots above depict the relationship between average swing path tilt and aSwing+, and average attack angle and aSwing+ among Major League Baseball hitters this season. Both of these scatter plots display a clear trend that swinging the bat on an upwards plane, and making contact when the bat is traveling upwards, is advantageous when attempting to generate power production. This finding confirms the intuition that Ted Williams displayed in The Science of Hitting that swinging on an upwards plane is ideal for hitting the ball in the air, and runs counter to the “old school” notion that fly balls are generated by swinging down to generate backspin. I find it particularly notable that all players with an average attack angle of 15 degrees or greater possess above-average aSwing+ values, and perhaps this could be a “magic number” for identifying above-average swings, similar to how 85+ MPH is a “magic number” for identifying above-average sliders (or this could simply be selection bias).
More bat speed is certainly better than less bat speed. Steeper swings are generally better than flatter swings. As shown by the plots above, the best swings in Major League Baseball combine both of these attributes, possessing above-average bat speed and either above-average swing path tilt or an above-average attack angle. These findings are represented on the Top 10 aSwing+ leaderboard displayed in the previous section, with each hitter present displaying top-tier bat speed AND a steep swing path tilt or attack angle. Swinging the bat fast and at an upward angle equals an increased probability of hitting the ball hard and in the air.
The scatter plot above depicts the relationship between average (estimated) swing acceleration and aSwing+ among Major League Baseball hitters this season. Estimated swing acceleration is a measure of a player’s ability to generate fast bat speeds with shorter swing lengths, which, in theory, provides the hitter with the ability to hit for both power (given the plus bat speed) and contact (given the shorter swing length). Displaying the ability to accelerate quickly is likely captured in the model, given the presence of both bat speed and swing length as features, and as the scatter plot displays, there is a slight, notable relationship between quicker acceleration and higher aSwing+ values.
Upon release of the latest batch of bat tracking metrics, ideal attack angle rate appeared to be the most confusing new metric, since while the metric made intuitive sense, it did not appear to correlate with offensive production when utilized in isolation. For reference, Statcast considers ideal attack angles to be between 5 and 20 degrees. As shown by the table above, aSwing+ is able to better explain the relationship between ideal attack angles and power output, as swings with an attack angle between 5 and 20 degrees generate significantly higher grades (with an average of 116 aSwing+), than swings that fall outside of this ideal attack angle range (avg. 88 aSwing+).
In my opinion, the most interesting finding of this research has been to discover how different pitch types affect the swing quality of offensive players, which can perhaps provide some insight into current pitching trends that are occurring throughout Major League Baseball. As shown by the table above, fastballs (four-seam, sinker, and cutter) generate the highest quality swings on average, with a 107 aSwing+, while breaking balls and off-speed pitches generate below-average aSwing+ values. This aligns with the observed trend that hitters generate more offensive production against fastballs, and is perhaps a main reason why teams have been dialing back fastball usage in recent seasons.
Discussion around the two-strike approach in hitting has increased in popularity in recent seasons, as the introduction of bat tracking data has allowed the public to analyze both whether the two-strike approach actually exists and whether it is the optimal strategy for hitters to utilize when they are behind in the count. I conducted a rudimentary analysis regarding the efficacy of the two-strike approach earlier this season, and Scott Powers presented on this topic at Saberseminar last season, which I highly recommend watching. As shown by the table above, it is clear that a two-strike approach exists as it pertains to swing quality, as the average aSwing+ value is clearly lower in two-strike counts, compared to early and leverage count states for the hitter.
The table above depicts the average aSwing+ value for each swing, grouped by pitch type AND count state. It is very evident from this table above that hitters take high-quality counts against fastballs when they are ahead in the count, with an average 118 aSwing+ against fastballs in this count state. Off-speed pitches generate below-average aSwing+ values in all count states, while the average quality of a hitter’s swings against breaking balls is dependent on the count state. As mentioned earlier, these findings are indicative of why teams have been dialing back fastball usage in recent seasons, and should give pitchers pause regarding using their fastball when they are behind in the count.
Not only is aSwing+ a powerful metric to utilize to evaluate a player’s raw power due to its predictive power, but it is also powerful due to its ability to become reliable in a small sample size. As shown by the table above, aSwing+ reaches an .80 level of reliability after about 50 swings, indicating that a player’s raw power ability can be reliably within a sample size of about 10 games. This is similar to Stuff+’s ability to provide valuable insights into a pitcher’s potential within a small sample size, and illustrates aSwing+’s potential as a predictive metric for analyzing offensive performance.
Moving Forward + Next Steps:
Overall, I am impressed with the performance of my swing quality model, aSwing+, and I believe that there is valuable use in utilizing the metric as both a tool for predictive analysis, as well as for contextualizing how the bat tracking metrics interact with each other to generate offensive production. That being said, aSwing+ is not a “one size fits all” metric for analyzing offensive production, similar to how Stuff+ is not the sole metric that should be used for analyzing pitchers. Possessing a high aSwing+ does not guarantee that a player will generate high levels of offensive production, and possessing a low aSwing+ does not mean that a player will generate no value offensively.
Making quality, “squared up” contact is still a vital element of hitting, and possessing a high aSwing+ is useless if a player is not able to make contact against upper-level competition. This is similar to how a pitcher with a high Stuff+ is unlikely to succeed long-term if they do not possess a baseline level of command ability. While bat tracking data is not publicly available for Minor League and College players, I would hypothesize that there are numerous players who possess high aSwing+ values at these levels that do not possess the minimum baseline of contact ability required to be productive Major League hitters. Wielding a high-quality swing that can generate power output is a very important piece of the puzzle to becoming a productive Major League hitter; however, it is critical to analyze this attribute within the context of a player’s swing decision and contact ability to grasp a better understanding of a player’s offensive potential.
While it has been emphasized numerous times throughout this article, aSwing+ is designed as a metric to analyze a player’s raw power ability, not as a proxy for a player’s overall level of offensive production. As discussed in further detail in the methodology article, I also attempted to create a “contact ability” swing path grade in addition to the “raw power” swing path grade (which became aSwing+); however, simply observing a player’s contact ability consistently outperformed a “contact quality” swing path grade. I believe that there is still potential for creating a contact-oriented aSwing+, however, it will likely require additional bat tracking metrics, such as hand position, or biomechanical variables such as spine angle, to better understand the precise path of the bat as it moves in space throughout the swing. While the existing metrics evidently provide enough information for quantifying a player’s raw power ability, these additional metrics would hopefully provide more insight into how quality contact is created, and likely improve aSwing+ in the process.
Since the implementation of Hawk-Eye into Major League Baseball stadiums in 2020, pitchers have consistently been the recipients of the valuable insights this tracking system has provided; however, with the release of bat tracking metrics, there is a sliver of optimism that this dynamic might tilt back in favor of the hitters. The development of swing quality models, such as aSwing+ and Jake DeBoever’s Swing+, are exciting initial steps in utilizing these new bat tracking metrics to better understand how offensive production is generated, and I will be releasing a couple more metrics quantifying offensive production using these metrics within the coming weeks. As we venture into the future, the integration and continued evolution of these bat tracking metrics stand poised to redefine how we analyze offensive production and further our understanding of the science of hitting.
Thanks for reading!
View the current aSwing+ leaderboards at the following link: 2025 aSwing+ Leaderboard
Follow @MLBDailyStats_ on X and Adam Salorio on Substack for more in-depth MLB analysis. Photo credits to Matt Dirksen/Getty Images.