# Bullpens, Tracking Elbow Torque, and mStress

This article builds on the previous research that we did with the MotusTHROW Sleeve that looked at the relationship between elbow stress and velocity.

Here, we will look at how those relationships have changed (or if they did) under a larger sample and discuss why normalizing stress with velocity is so important.

*(This article was written by Michael O’Connell, Research Assistant. Lightly edited for content by Kyle Boddy, Director of Research and Development.)*

**Updated Bullpen Numbers**

In our first go-round with the Motus Sleeve we originally had 17 athletes throw bullpens. It was later updated to 19, but one of the main points we saw was a smaller-than-expected relationship between velocity and stress, R^2 of .29.

Since those 19 bullpens, we’ve had 51 additional athletes throw high-effort bullpens for a total of 70 pitchers. We took the 6 fastest fastballs, for a total of 420 pitches, to use for our analysis. If a pitcher’s 6th pitch was tied by velocity, then we averaged those two or three pitches.

Here are the numbers from the original 19 pitchers.

Here are some updated numbers based on what we’ve found.

As always the numbers can be found here

Interestingly enough, the mean stress barely changed; it went up 0.8 Nm. The standard deviation of the stress decreased from 15.7 Nm to 12.8 Nm.

If we go back and compare the mStress measurements, we see that they are remarkably close as well.

The most interesting difference probably comes from the R^2 between the elbow-stress measurement and velocity. It decreased from 0.29 to 0.17.

So, there are really two ways you can look at the new Rˆ2 between stress and velocity.

First, you can believe that it’s a small difference, (since .17 of 1 is fairly small) and then you probably imagine that there are relatively small numbers of variables that account for velocity.

Put another way, if you believe that there are only, say, 5-8 main factors that could explain why pitchers throw hard, then, yes, velocity isn’t probably that much different than whatever those other variables are.

But, if there is a much larger number of variables that explain how a pitcher throws hard, say 15-20, then all of a sudden velocity seems much more important.

For example, Werner et al found 10 parameters related to pitching mechanics that accounted for 68% of the variance in ball velocity. Which would suggest a larger number of variables explain velocity.

We can expand our analysis to a multiple regression, including all of the metrics that Motus provides: Stress, Arm Speed, Arm Slot & Shoulder Rotation. After doing this, we end up with an R^2 of .23. Meaning the four metrics listed in the Motus app can explain 23% of the variance of fastball velocity.

So, adding these three variables only raises the R^2 0.06, from 0.17 to 0.23, which suggests that there are probably a larger number of variables that create velocity.

The variables that end up creating velocity are probably a combination of forces (such as elbow torque), mechanical differences (amount of horizontal abduction, elbow flexion, or knee flexion at stride foot contact), and minute-timing differences (such as decreased time to peak internal rotation). Some of which we can measure with the MotusTHROW, and others we need more equipment to measure.

Finding out which variables relate to faster pitching velocities is simply the first step. If (or when) all of those variables are discovered, then an entirely new process of discovery begins: how to safely make changes in those variables for athletes in order to increase velocity.

In future blog articles, we will further investigate the multiple factors related to velocity that are similar to the studies mentioned above.

**mStress: Normalized Stress & Velocity**

mStress = (Stress / Velocity) * 100

Normalizing velocity with elbow stress is going to make it much easier to compare pitchers across the spectrum to one another. mStress is the metric that we introduced in our offspeed article and the average from our pitchers was listed in the graphic above.

mStress helps put the numbers into a better context across different ages, and it will allow for better comparisons for pitchers who throw similar velocities

This is especially important if velocity is going to be one of the factors related to injury, then we should adjust for it.

Here we compare pitchers who threw 95+ with pitchers who threw 86-88.

Now, we can more clearly see that every pitcher is ** not **experiencing the same stress. Although some of the differences may seem small, they become more clear when comparing their mStress.

Training consideration

couldbe made based off of those numbers, butfirstfurther investigation would be needed to see what created these difference in the first place. Maybe some players have increased stress for mechanical, strength, or mobility reasons.

While being able to create and know a player’s mStress doesn’t give us answers, it allows us to ask much better questions and hopefully come closer to keeping players healthy.

As we’ve covered previously, this will also be very important when discussing offspeed pitches as well. Much of the peer-reviewed research has suggested that pitches, such as curveballs, are lower stress than fastballs. We’ve also found this to be the case in our own research.

But when stress becomes normalized for velocity, then the relationships flip. The curveball becomes much more stressful.

Looking at these numbers in a different context can adjust the meaning. The relationship we found with the curveball may be part of the reason for the why peer-reviewed research suggesting curveballs are less stressful while anecdotally athletes and parents find them more stressful or dangerous.

**Peak Stress Can Still Be Helpful**

There are a few caveats we have with using peak-stress measurements that we have mentioned before.

- Isn’t normalized per velocity
- Doesn’t account for muscles
- Doesn’t account for anatomical differences

That being said, with the growing popularity of the Motus Sleeve and baseball tech, it’s still useful. The important part is that knowing the caveats will give you a better understanding of what you do and don’t know.

If you are going to use the mTHROW during bullpens, then you have the ability to track changes that occur during that pen. Over the long term, using the Motus Sleeve gives you the ability to track an athlete’s workload.

Or if you are a coach with a few sensors and many pitchers, there is no reason why you can’t have athletes “check in” by wearing the sensor every few bullpens. A coach can then keep track if there are any small or big changes between bullpens.

There are multiple ways that you can use this technology to keep track of athletes’ workloads and any possible changes that are occurring.

Plus, it gives a general marker that is much better than having no data at all.

**We’ve published other articles summarizing our research, check them out here!**

We see that velocity does not always translate to higher stress on fastballs. But I am curious, has any research been done to determine stress levels of a cut fastball compared to a normal 4seam fastball or even compared to a curveball?

We haven’t looked into a cut fastball, yet. But we have looked into curveballs, sliders, and changeups.

https://www.drivelinebaseball.com/2017/02/fastballs-offspeed-pitches-comparative-relative-elbow-stress/

So in the study, pitcher 39 has a much higher stress level than pitcher 31. But because we cannot determine how that stress is “spread” over the different muscles, we cannot use that as an accurate injury predictor, correct? If that’s the case, what value does the Motus Sleeve have in terms of giving stress numbers?

That is correct. The literature on elbow stress and it’s relation to injury is currently only snapshots. One day of measurements.

The Motus sleeve can be used over time to see how (or if) stress numbers change and how those changes may or may not relate to injury

[…] we also wanted to see what would happen if we took velocity into account—mStress but for offspeed pitches. So we took the torque numbers and normalized them for their velocity to […]

If there is low correlation of stress and velocity I am not sure how dividing the Motus stress by velocity (Mstress) mathematically improves the correlation to draw better conclusion on crve ball, fast ball etc.