Injury Risk, Performance, and Velocity


It’s fairly common to watch a game these days, see a pitcher throw 99 mph, and hear the reaction of fans be a mix of amazement and fear: amazement that certain pitchers can throw that hard, and fear that it’s unsafe and an injury is guaranteed.

Given the many misconceptions and unknowns related to velocity, this article focuses on all things velocity by taking a deep dive into what the research says about injury risk and performance.

We first look at the history of studies that try to find risk factors for UCL tears and increased torque values in baseball pitching. We see that increased velocity is typically associated as a risk factor for injury, but there are still some gaps in the research that need to be filled in.

We then look at some new research that filled some of the previous research gaps. We take a wide look at what that means and what should be researched in the future.

Lastly, we discuss the performance benefits of throwing hard and look at how the research can be balanced with the performance benefits.

Velocity and Injury Risk

With shoulder injuries decreasing and elbow injuries increasing, many have turned to velocity as the culprit (Conte, Camp & Dines, 2016) because we have seen an increase in velocity in the major leagues over the last few seasons. Some pitchers may be throwing harder as they age, but teams are also replacing older pitchers (who throw at lower velocities) with younger players who throw harder and pitch in shorter spurts. These effects can be seen when looking at velocity aging curves and yearly velocity trends over time.

Given this perceived link between velocity and injury, there has been a lot of research looking at fastball velocity and injury rate. We’ve briefly summarized the research as much as possible and arranged the list in order of the year that each piece was published.

  • Olsen et al., (2006) sent surveys to adolescent pitchers who had shoulder or elbow surgery and found that athletes who threw faster than 85 mph were 2.58 times more likely to be injured. However, this was a survey, so these findings may be biased due to inaccurate self-reporting methods that potentially inflate velocity measurements.
  • Bushnell et al. (2010) followed 23 pitchers over 3 seasons and found that the 3 pitchers with the fastest velocity in their sample all required arm surgery. They saw a statistical difference between the non-injured velocity (85.22) and injured group velo (89.22), but they had a small sample size and only looked at the fastest pitch thrown (that is singular, the fastest single pitch) in one spring-training game over a 3-year period.
  • Whiteside et al. (2016) found six variables in their analysis that related to having UCL surgery: fewer days between consecutive games, smaller repertoire of pitches, less pronounced horizontal-release location, smaller stature, greater mean pitch speed, and greater mean pitch counts per game. They saw that a pitcher was 38% more likely to undergo elbow surgery for every 1 m/s (2.23 mph) increase in mean pitch speed. Somewhat confusingly, the authors also wrote that this finding is “an unlikely increase at the professional level,” even though their data set only looked at major league pitchers. The authors suggested the most logical explanation is that elbow experiences higher torques as velocity increases.
  • Chalmers et al. (2016) found higher pitch velocity was the most predictive factor of needing UCL surgery in MLB pitchers. Height, weight, and age were also found to be secondary predictors. Peak pitch velocity was significantly higher among pre-injury pitchers compared to controls (93.93 mph vs 92.1 mph). Mean pitch velocity was also higher in the pre-injury group (87.8 mph vs 86.9 mph).
  • Prodromo et al. (2016) compared 114 injured pitchers to 3780 matched controls (players of similar stature who were not injured) and found that pitchers who had higher velocities on a number of different pitch types—including fastballs, sliders, curveballs, changeups, and split-fingered— were at greater risk of needing UCL surgery in the future. The researchers found no significant difference in pitch selection between the two groups.
  • DeFroda et al. (2016) looked at UCL injuries from 2007 to 2014 and saw a statistically significant difference in the mean fastball velocity of pitchers who were hurt when compared to healthy pitchers (91.7 mph versus 91.0 mph).

Unlike the studies summarized above, which have all found links between higher velocities and inflated injury risks, Keller et al. (2016) found that MLB pitchers who needed UCL surgery did not pitch at higher velocities than matched controls. Instead, the researchers found that pitchers who threw a higher percentage of fastballs were at heightened risk for injury. While interesting, this finding is somewhat peculiar considering previous research and the fact that one often assumes that pitchers who throw harder also throw their fastballs more frequently.

Velocity and Increased Torque

An important aspect to note when reading the research above is that they are all correlating velocity to injury risk, but fail to consider torque. Many of the articles state, in some form or another, that velocity is likely linked to higher torques on the arm, which is why they see a relationship between velocity and injury.

While velocity isn’t the only factor, it is a theme that keeps coming up as a risk factor for injury. It may be easy to assume that higher velocities always mean higher elbow torques, but the relationship between the two hasn’t always been very clear.

There are also some confounding factors that need to be mentioned:

Selection bias is an issue. There isn’t a sample of pitchers with the same demands throwing in the mid 70s or mid 80s. The closest comparison would be high school and college athletes, but they are not as physically developed, and their seasons are more varied and shorter than the major league season.

Workload is another factor that comes to mind, as many of the studies listed above mention days between outings as a factor. DeFroda et al. found that more UCL injuries occurred earlier in the year and that there can be differences between starting and relieving.

These studies also largely looked at the averages, but it would be interesting to look at the distribution of injuries in larger velocity buckets as well. For example, would possible injury risk increase as velocity increased if buckets of 90-92, 92-94, 94-96 were compared? Or would we see that injury risk increases after a certain point, say 94 mph, and then level off even at increased velocities?

It’s also important to point out that the significance between velocities was relatively small—with the exception being Bushnell et al. (2010) which only looked at one pitch during a spring training start. (Pitchers can experience lower velocities in spring training as they build up work capacity for the season.)

  • 91.7 mph versus 91.0 mph
  • 92.08 mph versus 91.33 mph
  • 93.93 mph vs 92.1 mph
  • 89.22 mph versus 85.22 mph

Many studies also tend to look at raw torque numbers that are not normalized to height and weight, meaning taller pitchers will have higher torque values in part because their limbs are longer. There is also little research on what the muscles are doing during the throw. Since we’ve seen torques on the elbow can handle what cadavers are able to handle, the muscles of the arm play a key role in staying healthy.

There have been previous research studies published on both college and high school pitchers that have examined the specific relationship between velocity and elbow torque. Especially relevant with the increase in injuries at those levels, researchers have found a relationship between velocity and elbow torque, but have disagreed about the strength of the relationship.

For example, Hurd et al. (2012) found a positive association between ball velocity and elbow-varus torque (r^2 = 0.373, P<0.01) when looking at 26 high school pitchers. However, the average velocity of the pitchers within their sample was 71 mph, which is quite low.

Building off of these findings, Post et al. (2015) looked at a larger and older sample of 67 collegiate pitchers who averaged an 83.5 mph on their fastball, but didn’t find a significant correlation between ball velocity and elbow-varus torque.

A major reason why researchers have traditionally found such inconsistent results when attempting to link velocity, torque, and injury together is that their methods almost always look at group-wide averages and compare a wide range of pitchers. We believe that this causes them to overlook velocity’s relationship to torque at the individual level. How an individual experiences throwing harder is arguably more important than a group-level analysis, given that each pitcher has his own physiological attributes that influences results.

New research gives us a great look at how different the data can appear when comparing pitchers to each other instead of comparing them to themselves, which is why the recent paper Fastball Velocity and Elbow-Varus Torque in Professional Baseball Pitchers by Slowik et al. is so important.

New Research Fills in the Gaps

Slowik et al.’s data was collected via a retrospective review of ASMI’s database. That is, ASMI looked back through their database and included all pitchers who threw at least 5 fastballs and had a velocity range of 5 mph. To avoid outliers, no single pitch accounted for more than half the velocity range. They ended up analyzing 64 pitchers (52 righties and 12 lefties whose average velocity ranged from 71 to 96 mph) by normalizing elbow torque to body weight and height so that they could better compare the pitchers to each other.

They found that when they looked at the group of pitchers, velocity only explained 7.6% of variance in elbow torque. However, when they looked at individual pitchers, velocity explained 95.7% of the variance.

These findings can be summed up in two main ideas:

  1. If you compare multiple pitchers who throw the same mph, some will have higher torques than others. Not everyone who throws 92 mph is experiencing the same torque on their arm. Furthermore, one pitcher could throw 91 mph, another could throw 95 mph, and they both could experience the same elbow torque, on average. This clearly suggests that mechanics play a role in limiting elbow torque during a baseball pitch, but that topic is outside the scope of this post. This would also be part of the reason why the previously mentioned papers looking at torque and velocity didn’t see a standout relationship, as they were looking at the whole group and not at the individual level.

One interesting note about this finding is that we saw something similar years ago when looking at pitchers with the motus sleeve.

2. If you take one pitcher and have him throw harder, he will experience more torque on his elbow.

Of course, in Slowik et al.’s study, there were a few exceptions to this. While the full data set was not made available in the paper, the chart in the paper showing the relationships between normalized elbow torque and velocity saw some individuals stand out. A few pitchers saw no change in torque as they threw harder, while one or two actually saw a decrease as they threw harder.

In a perfect world, we would have this biomechanical data available to go with performance metrics to make personalized recommendations for pitchers. Since we don’t have that information available, we need to stay with the idea that the harder a pitcher throws, the more stress on his elbow he’ll experience. (We explain the nuance of that statement in the next section.)

So, when you’re watching a pitcher throw at the high end of his velocity spectrum, you can assume that he is experiencing higher levels of torque than usual. But that does not mean he is experiencing high levels of torque overall.

It’s unknown what intent level the pitchers were throwing at in Slowik et al. Although it may seem like a small difference, we aren’t sure if the results in this study would apply to pitchers throwing at max effort.

Drawing from Slowik et al., there are three ways that we could see results if we were only looking at pitchers throwing as hard as they could.

  1. Velocity and torque increase linearly, Which means, similar to this study, the harder a pitcher throws, the more torque he experiences on his arm.
  2. As velocity increases, torque increases exponentially, which means that after a certain point, pitchers that try to throw hard would see a higher increase in torque than would be seen linearly.
  3. At a certain point, velocity increases but there is no significant change in torque. This suggests that at a certain point, velocity can outpace torque increases at the very highest limits of velocity.

As mentioned above, all three of these examples were seen in individual pitcher comparisons. The question worth answering is which of the three theories best represents throwing at 90, 95, 100% intensity, or whether is it still a combination of the three, similar to the research above.

We can look elsewhere for hints. Research looking at wider levels of intent off the mound found fairly stark differences in how velocity and torque changed (Slenker et al., 2014).

In a similar theme as above, we saw that pitchers could drop 10 mph when comparing pitching to flat ground, but only see a decrease of 3 Nm of elbow torque when using the motus sleeve.

We’ve also seen pitchers increase velocity with no statistically significant increase in torque when they pulldown. Pulldowns are the closest thing we have to a maximum intent throw.

The differences in the stress metric between ASMI and the motus sleeve can be explained in our validation paper on the motus sleeve.

So we’ve seen that torques outpace intent level at low intent on the mound, that large decreases in velocity on flat-ground throws results in small decreases in torques, and that large increases in velocity while throwing with a running start results in small increases in torques. So, depending on the intent level and throwing drill, torque has been shown to have a nonlinear relationship with velocity.

Research looking specifically at pitchers throwing at the higher range of intent on the mound will give us an idea on how that torque changes.

However, regardless of the throwing modality, what we see is that velocity is a risk factor for injuries in professional pitchers because it is likely that the harder one pitcher throws, the more torque he experiences. Though, we don’t yet know whether the relationship between velocity and torque changes at the highest level of intent. Because ASMI looked back retroactively, and we see that athletes tend to throw slower when they are markered up, pitchers may have likely downregulated to throwing at a lower than game-like intensity.

So, this means that players who throw harder experience higher torques and that’s why they get injured, right? That’s a good guess, and it is definitely a factor, but there is some nuance to that statement.

Velocity, Torque, and Injury

The most interesting question after looking at this research is which matters more: the raw torque number or the relative torque number? Let’s elaborate.

There is a good amount of research pointing to higher velocities being a risk factor for injury. New research shows that it’s difficult to compare torque among pitchers, but as one pitcher throws harder, he’ll experience more stress. For the first point, we used the example that a pitcher throwing 91 mph could have the same stress as a pitcher throwing 95 mph. What also is true is you could have multiple pitchers throw 95 mph and have a variety of stresses.

Looking at Slowik et al., the mean value for normalized max elbow-varus torque was 5.33% ± 0.74% body weight x height. (The torque values were normalized to height and weight for better comparison.) The full data set isn’t available, but you can narrow in at a specific velocity and see a variety of torques.

For example, if we use 42 m/s (93 mph) as an example, we can see normalized torques at 4.7%, 4.8%, 5.4%, and 6.7%. This is similar, in theme, to what we saw using the motus sleeve a few years ago. Multiple pitchers can throw at the same velocity with different torques.

mStress is further explained here.

So, the question is: Does the pitcher who throws at 4.7% normalized torque have the same risk level as the pitcher throwing at 6.7% normalized torque? Does the pitcher throwing 93 mph at 60 Nm torque have the same injury risk as the pitcher throwing 93 mph at 85 Nm? This is important because we’ve seen many more correlations between velocity and injury than torque and injury.

We’ve seen velocity correlated to injury but we’ve also seen pitchers can have different torque levels at the same velocity, so what does that say about the relationship between injury and torque?

There is also more nuance involved than just torque measurements alone. Normalized torque may be better to compare than raw numbers, but we’ve also seen in several studies that workload, days rest, and throwing volume all play a role in influencing injury risk as well.

One study was able to link pitchers that experience higher levels of torque with increased likelihood of injury, but there were some serious limitations with the methods. The paper used hi-speed video cameras to get measure the biomechanics of 23 pitchers during one Spring Training game and then followed their injury history for three years after. The trend between elbow injury and higher torque was fairly significant (P= .0547), but they did see a significant correlation between elbow injury and higher elbow-valgus torque. The only issue with these findings is getting accurate torque measures from a multi-camera setup, and not a marker system, is incredibly challenging because it’s not yet been fully validated in baseball pitching.

For example, it’s difficult to find body landmarks while a player is wearing a uniform, and there is limited validation work between markered-based measurements and camera-based measurements for high-speed movements. It is speculated that a camera-based system could likely obtain accurate body positions at certain points, but it would have added difficulty calculating accurate torque measures when compared to markered-based labs because of an inability to measure rotation, and they can lose body positions as a uniform moves.

(Note: We run a marker based lab with Optitrack cameras and have recently partnered with KinaTrax markerless camera system to validate a markered-based lab to non-markered in order to see which measures are accurate.)

Therefore, we can conclude that velocity is still an injury risk factor and the harder one pitcher throws he’ll likely have higher elbow torque, but we still don’t know if higher normalized torque matters more or if pitchers are at similar risk levels throwing at their own higher torque. There is still more research needed in that area, especially for pitchers throwing at high intent. This is also looking strictly at velocity and not including mobility, strength, workload, and mechanics as other factors that could lead to higher risk of injury.

Given the link in research and general belief that throwing harder is more dangerous, it is often suggested by both fans and researchers that pitchers should just throw slower. We’re going to investigate whether that is a realistic strategy and what the consequences of decreasing velocity could be for a pitcher in terms of performance.

Velocity Performance Benefits

Again referencing the new ASMI study, it was recommended that pitchers should vary their velocities, because the more they try to throw at high intent, the higher torques they’re experiencing. Anecdotally, some hitters have said varying velocities makes hitting specific pitchers more difficult; however, the numbers behind this theory are lacking. You may also find pitchers who say they “vary” their velocities in the sense that they don’t throw at 100% effort all the time, but this strategy does not hold true for all pitchers, and there is no evidence that this strategy improves performance.

The ASMI authors supported their recommendations by looking at pitchers who qualified for the ERA title from 2015 to 2017 and the relationship between velocity and certain metrics. They found that higher velocities in pitchers resulted in lower seasonal outputs in ERA and WHIP, and subsequently higher seasonal outputs in fWAR and bWAR as well. With older, amateur research seeing similar relationships between velocity and K/9 in both starters and relievers, there’s evidence to support that velocity and performance are closely tied together.

However, the ASMI researchers instead concluded that velocity had a weak correlation with measures of performance, given that the r^2 values reported for each of the aforementioned performance metrics ranged from 0.034 to 0.158. These numbers contain a large amount of selection bias and would likely be different if relievers were included in some way. Requiring 162 innings means that there would be less variation in ERA or WAR because the pitchers who do throw that many innings are already very good or very lucky.

We also still have to remember that velocity is also treated as a floor, and further, we don’t know what exactly makes good pitchers successful in the first place.

Other academic research has examined what metrics correlate best to FIP. In a study by Whiteside et al. (2016) it was found that ball speed was one of three main factors in predicting FIP, but that it was also only able to explain 22% of the variance in FIP, suggesting that velocity is a big factor in performance, but there are still plenty of unknowns.

Interpreting how important velocity contributes to performance also comes down to differences in ERA, FIP, or whichever metrics you use to define “successful” performance in the first place. Since we don’t currently know of any model that can explain close to 100% of the variance in seasonal pitching performance, the scale is different than in other research. Complicating matter even more, metrics like ERA have low reliability year to year because they include a large amount of luck. That is why metrics like FIP, xFIP, and SIERA were created in an attempt to get a better understanding of how a pitcher is actually performing.

It would be different if we saw a larger range of fastball velocities in the major leagues, but we see few pitchers with fastballs below 90 mph. The pitchers who do fall below that range tend to be older and above average in other areas. The fact is that pitching is complex, and being a successful pitcher involves being good at multiple skills: throwing above the velocity floor, having adequate command, having at least one good secondary pitch (more than that for starters), and quality movement of all pitches, to name a few. Zack Greinke has even gone out of his way to show how important the velocity floor is for pitchers.

Greinke tweet

The possible benefits of not throwing at high velocity is often explained in a less than straightforward way, besides the supposed health benefits. General anecdotes (usually of pitchers who are older or already incredibly good) are used as examples that pitchers can be successful at lower velocities, which may broadly be true, but may not be true to you. The sample is also skewed since nobody sees pitchers with low velocity, bad command, and bad “stuff.” Those are often the first pitchers eliminated from the pool of possible pitchers.

Being successful at (relatively) lower velocities as a pitcher is less of a conscious choice and more of a series of “if, then” statements.

IF, you have above average fastball spin rate and high vertical movement…

IF, you have standout movement…

IF, you have above average control of multiple pitches…etc.

….THEN you can be successful at relatively lower velocities.

None of those mention possible mechanical differences of throwing at lower intent, which is something that may be further expanded on in further research.

This is in essence what Mike Fast wrote about on Hardball Times in his article “Lose a Tick, Gain a Tick in 2010. Pitchers who throw harder see a decrease in runs allowed per inning, and pitchers who throw on the lower end of the fastball velocity spectrum are often exceptional in other areas.

Asking pitchers to throw slower often means asking them to give away a competitive advantage for an unmeasurable and unknowable gain in health, which they might not be able to benefit from because throwing at lower velocity could mean worse statistics and a shorter career.

Many of the examples of pitchers having success with a lack of velocity can likely be attributed to confirmation bias and not because a statistical analysis showed an advantage. In fact, many front-office members would tell you that the injury issue is complicated and that velocity is one of, if not the most, important things for a pitcher.


Velocity is still seen as a huge benefactor, even when the straight correlations may be less than impressive. Giving hitters less time to react to a pitch is a good thing.


There was a lot to cover in this article, and while we know a bit more about the relationship between velocity and injury risk, there are still some unknowns. As always, it helps to look at all the things we do and don’t know.

What we know

  • Velocity is seen as a risk factor for injury, but it’s not the only one.
  • It’s assumed that velocity is a risk factor for injury because increased velocity comes with increased elbow torque
  • New research shows that pitchers who throw at similar velocities likely won’t be experiencing the same elbow torque (mechanics play a role).
  • New research also shows that as a pitcher throws harder, he’ll likely to be experiencing higher elbow torque.
  • Velocity is a beneficial piece to performing well.

What we don’t know: The exact relationship between injury and torque

That is, if three pitchers throw 95 mph with different torque, are they all at the same injury risk because of mph? Or are they different because of mechanics and other factors?

None of the research on velocity and injury risk makes velocity less important. Similar to other professional sports, baseball pitchers are pushing their bodies to their limits. They are going to incur a higher risk than an average Joe playing in a rec league.

This makes managing other factors so much more important because we want to get pitchers to throw hard while managing their risk. Managing mobility, strength and workload all matter greatly to every pitcher, especially to those who are either trying to maximize or maintain a high level of velocity.

Written by Technical Project Manager, Michael O’Connell

Tags for this post

Leave a reply