The Importance of Fastball Shape
Velocity is all the rage these days, and why shouldn’t it be? It’s fun to see that third digit light up on a radar gun or show up on the scoreboard. And it’s happening more often than ever. From Opening Day through the weekend, eight pitchers combined for a total of 28 pitches over 100 mph, with Emmanuel Clase hitting the mark nine times out of the 11 cutting fastballs he threw on Sunday. It’s certainly exciting, but velocity isn’t everything. Yes, I want to know how hard someone is throwing when evaluating a pitcher, but my first question after that is what is the shape of the pitch?
A decade ago, scouts based their fastball grades almost entirely on velocity. Above-average velocity? Above-average fastball. But with the emergence of technologies like TrackMan, Hawkeye and Rapsodo, that one-to-one relationship has become a relic. There are pitchers who throw in the upper 90s who have average fastballs because of their shape and other intangibles; there are some with average velocity that are nonetheless plus pitches for the same reason. Because of this, the scouting scale has changed and is beginning to capture variables outside of just miles per hour. Some teams have begun asking their scouts to grade fastballs across three traits — velocity, movement and command. When I ran pro scouting with the Astros, I asked our scouts to capture velocity in their reports, but I wanted their fastball grade to reflect the effectiveness of the pitch in a more holistic way.
The best way to learn about fastball shape is first to think about what constitutes a normal shape. Sixto Sánchez has some of the best velocity in baseball, averaging a remarkable 98.5 mph with his four-seam fastball in 2020. It’s a plus pitch to be sure, but it also doesn’t play like you’d expect from a heater thrown 98-99 mph. Among the pitches in his arsenal, it’s the third most-likely to put away an opposing hitter, and it’s where he gives up his home runs. Why? Because in terms of fastball shape, it’s exceptionally normal. Here are Sánchez’s four-seam fastballs in 2020, as measured by horizontal and vertical movement:

I added a “line of normality” to show just that, as the 45-degree angle shows the normal amount of vertical and corresponding horizontal break on a fastball. As you can see, Sánchez’s four-seamer has just a smidge more rise (vertical) than run (horizontal), but for the most part, the cluster of pitches sits right on that line. These pitches are moving the way most fastballs move. More importantly, these pitches are moving the way hitters expect them to when they come out of his hand. The end result? A pitch that is easier to hit.
So how do pitchers gain fastball effectiveness beyond what the radar gun says? By creating differentiation between their fastball and the norm by utilizing either rise or sink. The further one can get from the norm, the more effective the pitch can be beyond just the pure speed. Hitters expect a fastball to behave a certain way out of the hand and when it doesn’t, regardless of the direction in which it deviates, it creates challenges in terms of making hard contact, or any contact at all.
To illustrate, there was a recent four-year run with the Blue Jays in which Marco Estrada was a pretty darn good starting pitcher. His best year came in 2016, when he had a 3.48 ERA over 176 innings and posted 2.7 WAR. He did this while averaging just 88.1 mph on his fastball. Of course his exquisite changeup was his signature pitch, but he still threw his fastball nearly 50% of the time. There are plenty of minor league arms with plus changeups who are doomed for the waiver wire due to a lack of giddy-up on their heater, so how did Estrada survive and even thrive at times? Simple: he deviated from the norm. Here are Estrada’s four-seam fastballs from that 2016 campaign:

Look at that. It’s glorious. That is a fastball that isn’t behaving at all how you’d expect out of the hand. Hitters are going to be under that all day despite the below-average velocity. Estrada’s changeup was great, but without a fastball shape like this, he would never be able to set it up in the big leagues, as a normally-shaped 88 mph would get whacked around the ballpark before he got the opportunity to start changing speeds.
One of the current rising champs is Cleveland reliever James Karinchak who takes a fastball shape similar to Estrada’s and then adds to the nastiness with plus velocity by averaging in the mid-90s with the pitch. As you can see from his 2020 pitch chart, the shape is far from typical:

More akin to Estrada in the prospect world might be 2019 Milwaukee Brewers first-round pick Ethan Small, who sits with well below-average velocity (89-92 mph) from the left side, but has fastball shape similar to Estrada’s, allowing the pitch the play up significantly.
These rising pitches can easily be spotted with naked eye. Watch them ride through the zone without the normal downward plane towards the plate, and when they generate swings-and-misses, bats are almost always under the ball. This is also the best type of fastball to elevate with, creating further difficulty in raising one’s bat plane to even make contact. They don’t come without risks: by being in a higher than expected place when they arrive at the plate, contact tends to be of the fly ball variety, and with fly balls come home runs.
Pitch data is the primary reason for the rising fastball exploding in popularity over recent years, but teams have historically also valued those who instead of working above the line of normality, create pitches below it by utilizing sink.
The type of backspin and arm angle used on risers is far more dramatic than what is physically possible with sinkers, but the charts show how someone known for being a groundball machine can achieve the title. Here’s Jack Flaherty in 2020.

The cluster is not nearly as dramatically placed as some of the big risers, but it’s easy to look at the chart and figure out why it works for Flaherty. The result is either a whiff or (because of the lack of significant differentiation from normal) the bat hitting the top of the ball and driving it into the ground.
If you are looking for more dramatic differentiation, there is Aaron Bummer, a true turbo-sink southpaw who offers both real velocity and some of the strongest sinking action around:

That shape and 96 mph are unhittable at times, but as Bummer often shows, that combination of velocity and movement can also be difficult to keep in the strike zone. So while extreme pitch shapes can greatly improved overall pitch effectiveness, they often come with greater difficulty in terms of location.
If you are looking for something downright weird, there’s always Tyler Rogers, who can survive in the big leagues while sitting in the low-to-mid-80s by coming in from down-under and creating a shape that is literally almost off the charts; it has to be, considering the speed.

The downside of the sinker is that for most pitcher, there isn’t the same dramatic deviation from normal one finds with rising fastballs. The closer to normal one is, the more contact there is going to be, and when there is contact, bad things can happen. Because the pitch has significantly more horizontal movement than four-seamers do, that different-than-normal shape is essential, otherwise one is intentionally moving the ball into the natural bat path of opposite-handed hitters, creating sizable platoon splits.
While looking at these charts and thinking about each associated pitcher, you might have already come to an important conclusion. A big part of these shapes is created by arm angles. Guys like Estrada and Karinchak are coming from nearly over the top. Sixto Sánchez has a classic three-quarters release, and that 45 degree arm angle creates shapes that match the 45 degrees of the dreaded line of normality. Sinkers tend to come from an angle lower than that, or in the case of Rogers, as low as you can go.
Shape provides an essential part of the answer as to why fastballs play above or below the number on the gun, and teams are still working on discovering new advantages from these realizations. Some clubs have started to look at a second differentiation factor, one that examines the difference between actual and expected shape. Not from the line of normality this time, but rather from the shape expected from the individual pitcher’s arm angle. Others are beginning to theorize that a pitcher with a lower arm angle who can create some backspin and rise — as opposed to the expected sink — could suddenly be in possession of a highly effective fastball.
Shape is important, but like every aspect of the game, there are no absolutes, which is what makes baseball so great. There is deception and tunneling and sequencing, and then there are the fastballs that just confound what the data tells us. Take a look at Max Scherzer:

Scherzer throws hard. Not crazy hard, mind you, but in the mid-90s. Obviously his fastball is really good, but in terms of shape, it’s quite normal. I’ve spent considerable time contemplating Scherzer and I’m don’t have a clear answer for you as to how he bucks the clear and obvious trend. But what is life without a bit of mystery? Just enjoy the magic sometimes.
When it comes to the overall effectiveness of power pitches, velocity drives the bus, but pitchers can find greater potency on their fastballs by adding unique shapes. At the end of the day, when the catcher puts one finger down, it’s good to be a little different.
Kevin Goldstein is a National Writer at FanGraphs.
Scherzer hides the ball well in his wind up, so that would add “velocity” so to speak, in that the better only has X amount of time to pick up the ball from when he releases it. I’m sure how unpredictable a pitchers sequencing is when you have 3 to 4 plus pitches also changes how effective a fastball is, even if it’s a pitchers “worst” pitch (though I haven’t looked to see how Scherzer’s pitches rank, or looked in to sequencing etc etc). There’s definitely more to it than just the horizontal/vertical break, though I’m sure that raises the floor of each fastball so to speak!
Pivoting off your observation of Scherzer’s delivery, I’ve often wondered if guys who “show” the ball during the windup (eg, ball flashes behind head when the arm is beginning to come forward) are at any disadvantage, or if it really doesn’t matter?
I believe they are, but I’ve never seen data attached to it.
I think with everything there are too many variables than the raw numerical data (though I love that it gives a glimpse, and shows where pitchers can gain some edges for sure over others)
He’s not fooling anyone today 🙁
Thank you for this explainer! I really love FG for these types of more detailed looks at terminology and what exactly folks mean when they use certain phrases to describe pitches. Is it time for an update on pitch tunneling? Is that even still a thing?
ETA: Several FG notes on Sixto Sanchez have observed Miami seems to do well developing pitchers, so there’s no need to worry about his FB. How do we think the Marlins’s staff can “change” his FB shape?
Grips/finger pressure/seam-shifted wake or a different arm slot immediately come to mind.
I think the thing to keep in mind is that they may try some things (or may have tried some things), and it may also be that for his armslot and pitch arsenal, this above-average but fairly normal fastball is actually fine and better than any of the other outcomes. A pitch that is less ‘normal’ but that Sanchez can’t command as well may make him a less effective pitcher in aggregate, and Sanchez as a whole is a strong pitcher – its possible he can optimize, but it’s also possible that attempts to further optimize won’t produce significant improvement, and that that’s fine, because he’s already good.
The better the player overall, the more likely that attempts to tweak or change may be fruitless because the performance standard they have to reach to represent an improvement is higher. That doesn’t mean guys shouldn’t try, but it also doesn’t mean that guys who DONT improve aren’t trying: it may just be that their various attempts at improvement didn’t result in something that was better in the end, and so they were discarded.
This is really interesting.
Can we measure deceptiveness? Camera behind home plate, track the ball, see who hides the ball the longest before it comes out. Maybe calculate reaction time.
I think there are elements of deceptiveness beyond some easy to pick out things.
It’s fastball day on Fangraphs! Pairs nicely with the one on VAA.
Quite the debut for Justin! Really enjoyed that piece!
Having this article pop up right after Justin Choi’s on Jack Leiter is brilliant, in the full Tenth Doctor sense of the word. Someone please lock these two writers in a room and don’t let them out until they’ve produced a collaboration on fastballs.
The deception & the vertical movement was also what made prime Kershaw’s fastball so good too
Absolutely. His deception is hard to explain. I once described it as looking like the ball is coming out of his chest.
Like an Alien.
Peak Kershaw had like 99th percentile fastball rise to go with that tunneled slider.
Someone asked me a while back how/if possible it is to be a true 2-pitch MLB starting pitcher (absent Rich Hill and his like 5 variations on the curveball) and my answer was ‘imagine Clayton Kershaw without the curveball: not a HOFer, but still a good pitcher”
I feel like (this is just me talking out of my a**) that a lot of good deceptional deliveries & hitters that time these well would also be really good w/ syncopated rhythms in Jazz & in general in music. It’s all about the timing &taking away that timing – as someone who used to play instruments who was really bad w/ anything other than 4/4 time, I feel like there’s some link in your brain between the two lol
Regarding Scherzer and his fastball, I start thinking of someone like Mychal Givens, who throws from a low-3/4 or sidearm slot but actually has a normal-moving fastball. I recall someone describing that his movement doesn’t match his arm angle, as hitters would be expecting more sinking/riding fastballs from that slot. Perhaps Scherzer is similar, though I acknowledge his arm angle isn’t as low as Givens’.
Another pitcher off the top of my head with a similar quirk – Brian Moran. Throws sidearm but his fastball has a ton of cut.
“There are no absolutes”. Those four words define my problem with excessive use of analytics and I am a huge believer in analytics. I go back well before Bill James, to the days of Earnshaw Cook, and there is no question that much is gained from the information that is now available. I have wondered why baseball took 20 or 30 years longer than golf to realize the relationship between spin and launch angle. It was unbelievable that hitting coaches, even in the 21st century, were teaching to swing across the plane of the pitch instead of on the plane. I have been on this site for several years and the one overriding concept that I have seen is the belief, by those who believe in analytics, that they are absolutely infallible. If you want to get downvoted just try to make an argument that flies against that belief. What makes a good defender? The eye test tells us all we need to know. I should hope that a person working for a major league organization in a position to judge talent can look at quickness, agility, hand-eye coordination, arm strength, range and assimilate that into a better judgment than mere numbers. Those numbers can and should be there to polish the ideas but not dominate. Once players reach Double-A performance has to be the final determinator but if the player has out-performed his pre-determined projections he has to do it again and again to get a chance, but one of the selected ones gets an opportunity based on little more than where they were drafted. Just by chance I am watching Casey Mize climb the bump. If he wasn’t one of the selected ones, a Number 1 overall choice, his performance would not merit him being an the Tigers roster right now. In the meantime, let me sit back, enjoy the game and hope he pitches beautifully.
Its not that theyre infallible, but that they represent the best information we have and that they werent perfect so they were useless was the refrain of the antianalytics crowd back when they were reaching the mainstream.
Humans tend to focus their attention in biased and predictable ways. Outliers and memorable events have more
impact than they should. As you mention, an experienced talent evaluator may overcome some of this bias, but they’re still subject to it because they’re human and only see a small sample. Through supplementing their observational talents with statistics to identify situations where they often over or underestimate, they can become more accurate, as you see with, say, Longenhagen self-evaluating how he rates certain kinds of prospects.
But the assumption that the experienced person is just better than objective measurement is pretty much always wrong in basically every human field, because human beings are fallible and imprecise. We will notice things that systems and machines do not, but they do a far better job than we do of not getting hung up on the fallabilities and quirks of human memory processes, and can process large quantities of information far more efficiently than we ever can hope to.
The ole Derek Jeter defense discussion is a great example. Statistically, we know that for years and years, more balls hit towards Jeter went for hits than the average shortstop, but his defensive reputation was impeccable among evaluators of all stripes, because Jeter didn’t botch the balls he got to and often made acrobatic and impressive plays to convert them into outs. His defensive weakness was range, which just resulted in balls getting past that he could not reach. So he provided many memorable plays and very few botched errors to remember, but a broad look at the data shows how easy it was for everyone – including GG voters for many, many years – to consistently overlook his limited range because of how smooth and effortless he was on the balls he did reach. Their individual skill evaluations of Jeter were basically correct: Great hands, strong arm, clean exchange, less than normal range etc. The only thing that was wrong was their conclusion due to the relative value of those different skills not being what they thought they would be: they believed the skill and the arm and the sure hands made up for the range, but when looking at the aggregate results on grounders hit towards Jeter, we can see that they were wrong.
Now when you have evaluators who will take that data and use it to improve their future evaluation – to try and give more weight to the range element and allow their mis-judgment on Jeter to improve their evaluations of future players – you get better outcomes than either can provide.
I think Mike explains it quite well. As for the original post, I think Bosox’ first few sentences somewhat hit the nail on the head in a roundabout way. There likely would be an issue with “excessive” use of analytics (as “excessive” inherently implies a negative outcome). However, as Bosox point out, MLB’s relatively slow adoption of data like spin and launch angle could be used as evidence to show that baseball is actually not on the excessive end of data usage. Indeed quite the opposite could still possibly be true on aggregate. And the most important issues of the day, like player longevity, health, injury avoidance/detection, are still lacking in quantifiable and actionable data.
Like with many things, a combination of both well-understood hard data and good ol’ fashion human insight, will probably lead to the best outcomes. And I think we’re seeing that the best ball clubs nowadays are those that do well on both the number crunching and the human element (you can’t maximize your players’ talent if you are alienating them with lousy management!)
Also, “Fastball Shape”, does seem to me to veer more towards the “eye test” rather than abstract end of data (say like RPM). Humans can see ball movement, but cameras just see them more clearly, and spreadsheets help make sense of it all. But it’s definitely more visible and “human” than RPM which is near-invisible to the human eye and precedes the “human” things like the ballistics of the flight path of the baseball, which more directly affect outcomes.
As always: loving the discussions here!
You mention Marco Estrada and I will read the entire article.
https://blogs.fangraphs.com/the-rangers-worst-swings-against-marco-estrada/
What I like about Tyler Rogers is not only does he have a fastball that drops, he has a “rising” curveball to complement it. Such an interesting pitcher
I’ve heard several times of “bad” fastball guys. Kluber comes to mind as someone who had to overcome this. I’d love to hear if it’s his shape? Or a combination of the other intangibles mentioned? Is the plot data available via open info for me to peek by myself? 🙂
Available here at Fangraphs! Go to a pitcher’s Player Page and click on graphs –> Game Charts.
This is great stuff. I’m not clear, though, on why most pitchers are showing positive vertical movement. Don’t most pitchers release the ball higher than where it ends up?
https://tht.fangraphs.com/the-physics-of-a-rising-fastball/
The measured movement of a pitch is compared to a ball with zero spin. Fastballs have backspin and create “rise” in comparison to a ball with zero spin, which is displayed as positive vertical movement. Meanwhile curveballs have topspin and create “drop”, which give their pitches negative vertical movement. The higher the spin and spin efficiency, the more a pitch will work against gravity (positive vertical movement) or with gravity (negative vertical movement).
So basically if a pitcher threw a 90 knuckleball that ended up 2′ off the ground, if that pitch had 2000 rpm of backspin it might have had 12″ of positive movement and ended up 3′ above the ground. If it had 2500 rpm it might have had 18″ of positive movement and ended up 3.5′ above the ground. More spin means more movement compared to a ball with zero spin.
All pitchers except submarine pitchers release their pitches higher than where it ends up . The average MLB release height is 5.5′ – 6′, and pitches finish 2′ – 3′ above the ground. The “movement” of pitches are compared to balls with zero spin, but there are also stats that show movement compared to the average MLB pitcher. A pitch with18″ of movement may be only 4″ more compared to the MLB average though, since every pitcher throws their pitches with at least some amount of spin.
Thanks.
Scherzer possesses above average velocity and slightly above average rise from a well below average slot.
Really interesting! Always suspected that speed along wasn’t king (although it obviously helps if you can reduce the reaction time a hitter has).
Scherzer has a very low and horizontal release point, not as extreme as say Josh Hader, but it seems like low release points with a rising fastball leads to really good results.