Velocity in today’s game is certainly more central to the theme of the current pitching paradigm than ever before. More and more players are continuing to train by building up the capacities of their arm strength. Without this arm strength, a pitcher would be hard pressed to get an extended look from today’s talent evaluators. Players that possess raw tools will more often than not get the benefit of the doubt over a player that doesn’t possess the same tools but has all of the intangibles of an elite player. The philosophy is to gather up as many players possessing talent that “can’t” be taught and teach them the items that “can” be taught.
In pitching there is no more prized measurement industry-wide than velocity. Velocity showcases the combination of strength and efficiency and harder-throwing pitchers have continued to dominate the game. Higher velocities shorten the reaction time for a hitter to make a decision, in an instant where fractions of a second are the difference between a rollover and a ball in the seats.
What if the number that flashed on the radar gun was inaccurate? There’s a good chance it was, though not because the radar gun was faulty at reading the pitch’s velocity at release. In the context of the hitter/pitcher matchup, people are interested in a pitcher’s velocity because they want to know how much time the hitter has to react to a pitch. Reaction time is determined by the velocity and deceleration of the pitch over the distance the ball has to travel from release to the strike zone. There are two very similar metrics that toy with this idea that what we see may not truly be what we get.
One of them is known as effective velocity, coined by author and researcher Perry Husband. According to his website, effective velocity is “ a proven baseball training method using…metrics for pitcher deception and timing that applies to both pitchers and hitters in baseball.” Essentially it is a metric for timing that takes into account the reaction time for the optimal contact point on a particular pitch. However, that’s different from perceived velocity.
Perceived velocity is the attempt to determine the velocity of a pitch from the perspective of the hitter. Not all 95 mph fastballs are the same. Perceived velocity is a function of velocity and extension (the distance from the rubber that the pitcher releases the ball). When a player releases a pitch closer to the plate, that pitch will have a shorter distance to travel and hence will reach the plate sooner than another pitch of the same velocity further back. Due to a greater extension, the shorter distance will mean that the batter “perceives” the velocity to be faster than it actually is. In other words, a pitcher with a slower fastball can create the illusion of a “heavy” fastball by cutting down the distance between release and the hitting zone.
Perceived velocity can be utilized as a tool to deceive hitters that are expecting a faster or slower pitch upon release. If a hitter is guessing and happens to guess correctly on which pitch is being thrown, perceived velocity can be used to throw off the timing of the hitter. While it’s not always the prime indicator of pitching success, it can be an important factor in deception.
How is it calculated? Perceived velocity is a function of the pitch’s velocity and the extension of the pitcher at release. If the pitcher has a greater-than-average extension, that pitch will seem faster to a hitter. If the extension is below average, the velocity will seem slower. Perceived velocity is dependent upon how much ground the pitcher can cover and how significantly he can minimize the reaction time of the hitter. That is also dependent upon the size of the pitcher because the longer a pitcher’s limbs, the better perceived velocity we can expect from him. The research data from the 2020 MLB regular season was scraped from Baseball Savant for this report using the baseballr package.
There were slight variations in velocity additions by extension among pitch types. The extension played a far more significant role in adding perceived velocity to higher velocity pitch types as well as pitch types with lower gyro degree. Every extra foot of extension on a Major League fastball added, on average, 1.67 mph to the velocity separation.
Velocity separation, when mentioned in this writing, is referring to the difference between perceived velocity and release velocity. Positive velocity separation for a pitch means that pitch was perceived faster by the hitter, negative separation was perceived as being slower.
In the chart above you can see that the velocity separation of some pitches was more influenced by extension than others. Each number signifies the effect that an extra foot of extension had on the perceived velocity. If the release velocity remained constant, the extra foot of extension would’ve increased the perceived velocity, and hence, the velocity separation by the same amount.
It’s fascinating to note that upon reviewing the entirety of the data for this article, only 4.8% of pitches in 2020 were perceived equal to their release velocity. This means about 1 out of every 20 pitches demonstrated zero velocity separation.
Every pitch is different in terms of perceived velocity. The following discusses which pitchers best exploited perceived velocity separation and important pitch-specific factors.
The four-seam fastball is the pitch-type most influenced in perceived velocity by added extension. Fastballs also have the greatest variance in terms of velocity separation.
One of the best perceived velocity fastballs came from Tyler Glasnow of the Rays. He threw 584 of them during the 2020 regular season. His 6’8” frame also produced the longest average extension on fastballs in all of Major League Baseball (7.66 feet). If you also factor in the 2400 rpm spin rate, his average 96.9 mph fastball appeared to the hitter to have the same reaction time as a 99.4 mph fastball! This is one of the main reasons he has been so dominant. The ability to cut down the time that the hitter can gauge a pitch coming that fast makes it incredibly difficult to hit.
Another great example of this principle was the emergence of Blue Jays’ reliever Jordan Romano in 2020. The year prior, the distance of his extension was 7.41 feet and this year he added almost an extra two inches to his average extension. In addition, he made significant offseason improvements in arm strength. From this, his average 92.9 perceived velocity fastball in 2019 evolved into a perceived velocity of 98.4 mph in 2020. Against his 95 fastballs in 2020, hitters produced an expected slash line of .090/.173/.137 (xBA/xwOBACON/xSLG). That’s pretty impressive for a pitch he used almost 20% less than his slider.
However, harder pitching doesn’t necessarily translate into perceived fastball separation. Guys with lesser velocity like Steve Cishek and Brent Suter were still able to be effective in speeding up their fastballs to go along with their unusual deliveries. Suter, in fact, used his 85.3 mph fastball 71.5% of the time.
Some pitchers didn’t quite have the same reach upon release, creating a perceived velocity that was slower than their release velocity. The widest negative separation (-2.4 mph) came from the Rangers’ Jimmy Herget who was the only pitcher with an average fastball extension under 5.5 feet in length. He’s an anomaly in more ways than one. For someone with such a short stride, the height of his release was also very peculiar. It too was fairly low. That combination of release height and extension leaves his fastball an average 55.75 feet to travel to the front of the plate. That’s four inches behind the next closest pitcher. Did being such an outlier lead him to success last season? He finished with a 3.20 ERA in 19.2 innings, but was hurt in the categories of his own doing. He had a 6.4 BB/9 producing an awful 5.22 FIP.
Two-seam fastballs and sinkers were lumped into the category of fastballs with more exaggerated arm-side run. Just as in the four-seam variety, Steve Cishek and Jimmy Herget took top and bottom honors. One very notable name at the top of the velocity separation leaderboard was Kenley Jansen. Obviously, he’s a guy that relied a lot on his cut fastball, but his antithesis-breaking pitch paired to cover 90% of his pitches. While there’s very little actual velocity difference between his cutter and sinker, he got tremendous separation between perceived velocity and release velocity. He’s a large human being with long limbs and a long stride. Being as tall as he is allowed him to create more vertical break on his sinker (1.5 feet) than any other pitcher with elite velo separation on his sinker.
Codi Heuer. If that’s a name you’re seeing for the first time, you need to check out the video above. His stuff is electric. He was a rookie relief pitcher for the White Sox this past year. His arm action is a little funky, but he posted a top five velo separation on his sinker. To go along with nearly 2400 spin, the average perceived velocity on his sinker was 99.2 mph! He utilized his sinker 63.1% of the time and it was very effective, surrendering only a .278 wOBA.
For splitters, Statcast said Aroldis Chapman had the greatest velo separation. However, he only threw three of them all year. Right behind him was none other than Braves pitcher Touki Toussaint who actually had a terrible year to put it lightly. Five of his seven appearances were starts and he ended up with an 8.88 ERA and a 7.05 FIP. While his splitter was perceived to be faster than it actually was, it actually did not break much vertically. It only averaged about two inches of vertical break. And despite his terrible year, the splitter actually seemed to work very well for him. It was his best pitch in limiting almost every single offensive category. Perhaps most importantly, it prevented the gopher ball. He gave up 7 home runs in 2020: 2 off fastballs, 2 off curves, 3 off sliders, NONE off splitters. He tried adding a slider into his repertoire this year, but that backfired on him. The slider did not work for him because the velocity profile was too much like his splitter and all the while had far less horizontal movement than the splitter.
On the flip side, one player just claimed off waivers by the Mets was Nick Tropeano. He is the only player in Major League Baseball that has a splitter with an average perceived velocity below 80 mph. This equated to hitters seeing his splitter 1.08 mph slower than reality, by far the most for this pitch type. This was one of the reasons why he was exceptional, because opposing hitters had a tough time against pitches that he didn’t throw hard. He was excellent for the Pirates over seven games throwing his offspeed pitches 70% of the time.
An initial impression of the changeup was that a negative velocity separation would be more beneficial for a pitch that ought to be slower than it appears. For the most part that appears to be true; xBA and xwOBACON among elite negative velo separations are far lower than those on the opposite side.
Brad Brach, Mike Soroka, and Jesus Luzardo were three pitchers whose changeups appeared to be slower than reality. It’s interesting that 6’6” Brach and 6’5” Soroka had extensions almost a foot less than the ML changeup average, causing their velo separations to be negative. Brach and Soroka had low SLG and wOBA numbers against, but Luzardo had a higher whiff rate on his change. Soroka in his abbreviated season only used his changeup 12.1% of the time but mostly to put hitters away, striking out batters 60% of the time it was used with two strikes.
The high positive velo separation changeups yielded more mixed results. Some of the pitchers with high changeup velo separation like Glasnow, Heuer, and Luis Patino were more varied in their results. Glasnow only threw his changeup 4.7% of the time, but it got absolutely rocked when he threw it. It could’ve been because the perceived velocity on his changeup was only 0.06 mph off the average Major League fastball. Heuer threw his 9.4% of the time and while it didn’t generate a lot of swing-and-miss, it had really stellar results. There wasn’t a single hard hit ball off his changeup all year. Finally, Patino had pretty good expected stats, but was unlucky when his actual stats did not play out how he hoped, culminating in average results. Those differences were 87 points in BA, 192 points in SLG, and 101 points in wOBA, all working against him. For changeups, velocity separation was a better indicator of consistent success the more negative it becomes.
For the slider, Codi Heuer’s separation was significantly better than everyone else’s. What’s amazing about it is that he threw 93 sliders and hitters whiffed on two of every three swings they took against it. The slider is also Steve Cishek’s best pitch, which he threw 51% of the time. The average straight-line distance his pitch had to travel to the plate was shorter than any other slider in the game. He’s one of only 4 pitchers in MLB whose slider averaged more than a foot of horizontal break and owned an average extension over 7 feet. With this rare, competitive profile he ended with great opponents’ percentages.
A couple of names that had poor velocity separation on sliders included Jimmy Herget again, and Clarke Schmidt of the Yankees. Pitchers like these at the low end of the separation spectrum had sliders with elite spin rate. For instance, Clarke Schmidt’s slider averaged 3085 rpm. It’s tough to say what portion of their success was attributed to their velo separations and what portion was dedicated to a high spin rate.
Both Kenley Jansen and Brandon Workman had similar cutters in terms of velocity separations. On a pitch like the cutter that relies on late armside movement, it definitely worked in the pitcher’s favor to have a positive velocity differential. Squaring up a good cutter is one of the harder feats for a hitter to accomplish. Just ask Mariano Rivera, who made an entire Hall of Fame career out of throwing it. The best cutter in today’s game, Kenley Jansen’s, had the greatest positive separation of any cutter. One advantage is that he released the ball almost a foot ahead of every pitcher that used their cutter as much or more than him. In those 53.4 feet after Jansen’s release, it’s tough to center that pitch because it had no remote replica. You will not find a pitch at 90 mph and 2500 rpm that breaks 8 inches laterally and also 13.5 inches vertically. That combination of metrics is so unique to Jansen that it made his cutter appear 1.74 mph faster. It’s no wonder that the hard hit percentage on his cutter was only 13.9%. That’s about one in every seven balls put in play…not bad.
When it came to curveballs, Freddy Peralta was better than everybody else at making them look fast. While he’s only 5’11” tall, he was able to generate the extension of someone who is much taller than him. How about this for a stat? On average, Freddy Peralta released his curveball at the same extension as 6’8” Tyler Glasnow (7.32 feet), but Glasnow released his 14 inches higher. Also an interesting contrast is where we see Peralta’s curveball, having almost 17 inches of horizontal break versus Glasnow’s curve that breaks downward more at about 18 inches.
Perhaps the most critical factor in evaluating a curveball was the spin rate. When running a correlation matrix on curveball metrics, we got a correlation coefficient of r = -0.5 between spin rate and velo separation. That means as the spin rate goes up, the velo separation trends downward and vice versa. If you’re sitting there questioning the relationship between these two variables, then think about spin rate and its relationship to vertical break. Where would it fall on a scale of -1 to 1? The truth is that it’s also r = -0.5. As spin rate went up, the vertical break trends within the data indicated a moderate, negative relationship where downward movement increased. These two relationships exist at about the same strength level of correlation. With the exception of Jacob deGrom, the top 10 curveballs by velo separation had spin rates ranging from 2000 to 2400. In the bottom 10 by velo separation (with exceptions of Drew Pomeranz and Jace Fry) spin rates ranged from 2400 to 3200.
That’s not to say you can’t have success without high spin in the lower ranges of velo separation. I will refer back to the exception I made of Drew Pomeranz. He had 2100 spin on his curveball, which is low for a Major League curveball. However, the expected stats that it produced were fantastic: .134/.127/.146 (xBA/xwOBACON/xSLG). The relationship between curveballs and perceived velocity separation isn’t perfect, but, unlike many other pitches, it seems as though there is some kind of general pattern occurring.
In the chart above you can see an aggregation of almost every single pitch put into play this season. Clearly, any velocity separation made a difference. Hitters did better when the pitch approached exactly as they perceived it. Positive and negative separations both produced nearly identical average exit velocities. However, pitches with positive velocity separation tended to “beat” hitters causing them to take weak, defensive swings producing ground balls. That explains the difference in launch angle and distance between positive and negative separations. One thing is for sure, the evidence says it’s beneficial to have some amount of velocity separation. It could be positive. It could be negative. The more separation a pitcher had on either side, the better chance he had of limiting offensive production.
In 2020, some pitchers found success using perceived velocity to their advantage. Like many other factors that play out in our game, it all depends on the pitcher. What can be said about perceived velocity is that it does have an effect on the outcomes. Being able to create the disconnect between what the batter sees and what actually plays out is shown to create a distinct advantage for the pitcher.