As the analytical side of baseball becomes more accessible, so to have Baseball Savant’s gamefeeds. It adds another level of depth to the devoted fan’s watching experience. Along with asking how many runs a pitcher gave up, they could assess their pitch locations. Along with watching a ball soar into the seats, they could look at its data off the bat.
At each gamefeed’s window display are the top five in four metrics: exit velocity, distance, pitch velocity and pitcher whiffs, all in descending order. What’ll be directing the most foot traffic varies by game, but exit velocity is always a safe bet. After all, everyone loves admiring the sound of the ball off the bat. It’s a reliable quirk for each game and always leaves us wondering if we’ll see a historic feat or a noisy show.
Not much credence is given to this in overarching analyses, which is valid. It’s just a single-game metric that doesn’t compound over time. Even so, it’s fascinating to think about its effect on the outcome of the game. Or the implications of the outcome on how hard the hitters were swinging. So, to exacerbate an intriguing but bombastic metric, I decided to dive into it a little deeper.
Before I begin, I feel an obligation to acknowledge the primary flaws in this analysis. For one, the sample size used was only during the 2021 season. While that’s almost 5,000 games worth of data, there could be variance left to be settled. But of those 4,858 games, the data accounted for about 99.8 percent. The missing games likely will be a grain of sand on the beach in this case, but it’s worth mentioning nonetheless.
One thing that’s persistent throughout the visualizations for this analysis is the symmetry. This is because a single game will be included twice, one from the home team’s perspective and the other away. And one’s team split will be the other team’s opposite. While it won’t affect the integrity of the data, it may seem disorienting or overly coincidental.
To begin, I examined the distribution of the top five exit velocities by each team through a simple histogram. This would allow me to determine the probabilities of a certain exit velocity split (like 3-2 or 4-1) for the 2021 season.
Unsurprisingly, the chart shows that most teams had a 3-2 or 2-3 split in their top exit velocities downward. In 2021, about 57.2 percent of games had this split, followed by 34.2 percent for 4-1 or 1-4 splits and 8.7 percent for the lopsided slugfests. When looking at which teams had the most 5-0 splits, the New York Yankees top the list with 18, over 10 percent of their games.
The baseball community has come to know exit velocity as a valuable metric since its public implementation in 2015. Of the 14 MVPs in that timespan (all hitters), only one had an average exit velocity below the 50th percentile (2017 José Altuve) and nine had one at or above the 90th percentile. This power surge hasn’t been without its shortcomings, but there’s a general consensus that it’s a massive contributor to winning baseball games.
Given this nugget, my next objective was to determine how Savant’s gamefeed exit velocity correlated to scoring. To examine this, I paired the exit velocity splits with run differential so as to account for both great hitting and poor pitching. I used a box-and-whiskers plot to visualize this, with a dashed line at zero difference in runs.
Even without any numbers in front of us, it’s easy to visualize how hitting the ball harder than the other team in a game will lead to more success. But for the sake of it, I charted five different percentile intervals for run differential to quantify exactly how successful it is.
|Exit Velocity Split||10th Percentile||25th Percentile||50th Percentile||75th Percentile||90th Percentile|
I immediately noticed that, as a result of the aforementioned symmetry, when the splits flipped the percentile intervals mirrored each other. For instance, a team that occupies all of the five top exit velocity spots will outscore the other team by at least nine runs a tenth of the time. But a team that doesn’t occupy any of the five top spots will be outscored by at least nine runs about a tenth of the time. It’s wonky, but it distinctly shows how much of a difference jumping on the ball can make.
But how does it correlate to winning? Run differential is useful of course, but the “W” column won’t prejudice between outscoring your opponent by one or 20 runs. In games where a team had the majority of the top exit velocity split, they won 1,485 games, good for a 61.2 winning percentage. On the flip side, teams that had the minority won 940 games for a 38.8 winning percentage.
When examining those winning percentages over a full major league season, that’s 99 wins for the majority and 62 wins for the minority. Or, in simpler terms, it’s the difference between a division winner and the first overall draft pick.
Given the gamefeed’s correlation to a player’s power, there are certain players more apt at occupying top spots than others. Last year, the “top exit velocity in a game” king was Vladimir Guerrero Jr., who was also the runner-up for the American League MVP. He occupied that spot 59 times, just over a third of his games played. Not just that, but he also occupied the second and fourth spots the most.
Finishing behind Guerrero Jr. for the top spot was Giancarlo Stanton with 54, American League MVP Shohei Ohtani with 41, Manny Machado with 39 and New York counterparts Pete Alonso and Aaron Judge tied with 38.
During the baseball season, fans tend to gravitate to game-by-game analysis. As the community is treated to one contest per day, they’ll try to squeeze all the information they can out of it. So ultimately, the intrinsic value of Baseball Savant’s gamefeeds comes from that flash occurring 162 times a year. But even though it’ll be sparsely revisited by just the next day, that doesn’t mean there isn’t meaningful context to be found around it. And because of it.