Dr. Glenn Fleisig and his research group at ASMI published a study—as we have at Driveline—investigating the effects of weighted balls on pitching mechanics using optical motion capture (the gold-standard in biomechanics measurement).
The ASMI and Driveline studies both investigated how pitchers’ mechanics changed as they threw differently weighted balls, and both studies found similar changes. Both studies found that peak rotational speeds decreased, torso forward tilt increased, and pitch velocity decreased as ball weight increased from 3 ounces to 7 ounces, while the ASMI study produced a few additional results (although they didn’t adjust for multiple comparisons). One significant problem with these studies, however, is that they compared mechanics by measuring a few handfuls of single values for each throw, rather than comparing measurements throughout the whole throw.
These singular values to describe the biomechanics of a throw are calculated by taking various measurements at standard points in time. For example, measuring pelvis position at the time when the pitcher’s front foot strikes the ground (foot plant, FP) or when the ball is released from the hand (ball release, BR). We can also take measurements when they reach their global maximums or minimums, such as maximum elbow flexion, throughout the throw. However, there is much more to the pitching delivery than a few instantaneous positions, so being able to compare metrics for the entire duration of the throw is invaluable. To address this limitation, we used statistical parametric mapping to compare the entire throw using the same dataset as our pre-printed study. Here is an example with forward trunk tilt: Would you rather compare trunk tilt throughout two whole throws like the plot here on the left, or compare the degree of trunk tilt in these throws at one time point, as seen in the plot on the right?
We discuss this limitation in more detail in our first full signal analysis post with an example of our solution, but, in summary, full signal analysis using statistical parametric mapping (SPM) allows us to compare entire throws from start to finish, rather than at only a handful of time points. We should also mention that full signal analysis has its own limitations due to a large number of comparisons and because the signals need to be time normalized, but these are overshadowed by the benefits of full signal analysis.
Here’s a quick summary of the results of the study using full signal analysis:
- The arm moves faster with lighter balls and slower with heavier balls in terms of elbow extension and shoulder internal rotation.
- Lighter balls are associated with a bit more maximum elbow flexion than heavier balls are, but not much.
- Athletes reach a greater degree of shoulder internal rotation just before ball release with lighter balls, but not by much.
- Heavier ball throws are associated with more forward trunk tilt just before ball release (chest more parallel with the ground), but not by much.
- Athletes begin to resist external rotation earlier when throwing heavy balls than when throwing light balls, but not by much.
All things considered, the one major effect of 3 – 7 oz weighted baseballs is the speed at which athletes deploy their arms. Throwing these balls resulted in little to no ‘practically significant’ differences in pitching mechanics. This supports using weighted balls to possibly train more efficiently, improve athlete comfort, and train for power and speed.
And now for the specifics.
For this investigation we use the same statistics package as in the aforementioned analysis (statistical parametric mapping) to compare mechanics from the start of the throw to the finish. For example, one of the metrics we investigated was elbow flexion/extension velocity. The effect of ball weight on elbow flexion was evaluated for the whole throw using an ANOVA statistical test, and since the test showed that ball weight affected elbow flexion, we compared elbow flexion between pitches with different weights (3 oz vs. 4 oz, 4 oz vs. 6 oz, etc.).
Statistical difference between conditions is shown by the red line on the right plot; when it exceeds the threshold of significance (black horizontal dashed line on the right hand side), we can say with confidence that the two conditions are statistically different. For example, the red line exceeds the horizontal dashed line near the end of the plot below, telling us that ball weight significantly affected this metric near the end of the throw, since we know that the ball is released at frame 300 on the x-axis.
This process is basically repeated for all chosen metrics to see not only what aspects of the throw are different when throwing different ball weights but when in the throw they are different.
For the metrics that are shown to be significantly affected by ball weight, we compare each ball weight to see which are actually different. You can see the results of this type of comparison below, which compares throws with the 3 oz weighted baseball to throws with the 7 oz weighted baseball.
Similar to our investigation using point of interest data (e.g. comparing pelvis angle at front foot plant between 3 oz and 4 oz throws), we grouped biomechanical metrics into categories to describe overall types of movements in the throw. These categories were arm position measurements such as elbow flexion, mid-section position measurements such as pelvis rotation angle, lead knee flexion, the only lower body measurement we took, kinematic velocity measurements such as torso rotation velocity, and kinetics such as elbow valgus torque. The metrics that were significantly affected by ball weight within subjects (which mostly coincided with our point of interest analysis) included:
- Arm Kinematics
- Elbow Flexion/Extension
- Shoulder Rotation
- Mid-Section Kinematics
- Anterior/Posterior Trunk Tilt
- Kinematic Velocities
- Elbow Flexion/Extension Velocity
- Shoulder Rotational Velocity
- Arm Joint Kinetics
- Elbow Valgus Torque
- Shoulder Rotation Torque
This means that within each subject (each participant in the study compared to himself), the above metrics had at least one point in time throughout the throw which differed when throwing different ball weights.
Now let’s look further at each of those kinematic and kinetic changes.
Elbow flexion and shoulder rotation were both significantly affected by ball weight in the arm kinematics category. The two other metrics tested in the arm kinematics category, shoulder horizontal abduction and shoulder abduction, did not show a significant effect from weighted balls.
We will start with elbow flexion:
The red, vertical lines in the plots above represent the moment of foot plant. You can see that the black line on the right plot reaches statistical significance (by crossing the horizontal dashed line) just a bit after foot plant meaning that the two conditions become different at this point in the throw. If you look at the same spot on the left graph, you can see that the difference is around the peak of the elbow flexion signal with the 3 oz line being higher than the 6 oz line, meaning 3 oz throws resulted in a significantly higher peak elbow flexion on average. Higher peak elbow flexion corresponds to the arm being more bent at the elbow—their hand being closer to their body. This is what we found in our point of interest analysis as well.
Now for shoulder rotation:
This 4 oz vs. 7 oz comparison shows how shoulder rotation is affected by weighted balls, but it’s not where you might think. Maximum shoulder external rotation is not affected by weighted baseballs (at least 3 oz – 7oz) in a max intent pitch, but lighter balls have more internal rotation at the end of the throw, right before ball release.
This makes sense for a few reasons. First, shoulder internal rotation velocity is faster with light balls so it makes sense that light balls would have more internal rotation later in the throw. Second, the lighter ball weights may require a lower trajectory out of the hand, which could affect shoulder positioning at ball release. Last but not least, the position of the torso at ball release is slightly different, which will affect shoulder rotation since shoulder angles are calculated as the position of the humerus (upper arm) with respect to the position of the torso. More below.
To test mid-section mechanics, we included comparisons of pelvis and trunk (torso) forward and lateral tilt, pelvis and torso rotation, and hip shoulder separation. Only trunk forward tilt was significantly affected by weighted balls, as we see here.
At the end of the throw, athletes exhibited significantly more forward trunk tilt (chest more parallel with the ground) with heavier balls just before ball release, which we also found in our point of interest analysis. This is interesting in its own right because it suggests that athletes more actively use their mid-sections to throw heavier balls, but it’s also interesting because it ties in to shoulder rotation. Since shoulder angles are calculated by comparing the position of the humerus (upper arm) relative to the torso (the chest), if the torso is tilted further forward when throwing heavier balls but the arm is at a similar position at ball release, shoulder external rotation is going to be greater.
Shoulder internal rotation velocity and elbow extension velocity were both affected by weighted balls, but pelvis and trunk rotational velocity and front knee extension velocity were unaffected.
Light balls measured in terms of both elbow extension velocity and shoulder internal rotation velocity elicited higher peaks, meaning athletes extended their elbow faster and internally rotated their shoulder faster when throwing lighter balls. This is somewhat intuitive and also what we found with our point of interest analysis.
The only two kinetics tested in this analysis were the elbow varus moment (can be thought of as the same thing as valgus torque) and shoulder internal rotation moment (the torque involved in the internal rotation movement of the shoulder). In our point of interest analysis, we did not find a significant difference in peak values of either of these metrics, but we wanted to look at more than just the peak values in this case. The results are as follows:
We found (again) that peak values in both metrics of stress/kinetics/torques (whichever you prefer to call it) were not affected by weighted balls on average. We did find a significant difference between light and heavy balls at foot plant, but this difference is not practically important because it happens when the torques are right around zero. You can see that the 3 oz throws in both elbow varus moment and shoulder internal rotation moment produce significantly lower values than the 6 oz at foot plant, but it is a difference of 7.16 Nm for 3 oz and 13.04 Nm for 6 oz.
Since both of these values are so low compared to their respective peaks (about 147 Nm for both), the effect we see here of weighted ball on shoulder and elbow torques is not an implication for injury; rather, this effect has more implications on timing analysis, i.e. suggesting that athletes start to resist external rotation a short time earlier with heavy balls than with light balls.
What We Learned about Pitching with Weighted Balls
After analyzing commonly used discrete points of interest throughout the throw, incorporating previously published literature, and also doing full signal kinematic analysis on our current dataset, there is one thing we have learned about 3 – 7 oz leather weighted baseballs: they don’t change pitching mechanics hardly at all. Athletes move their arms faster with lighter balls and slower with heavier balls, but as far as positioning and movement effects, they are pretty minimal.
Sure, when you add a movement constraint into the mix like we do here at Driveline, the combination of a weighted implement and the constraint may affect how an athlete moves. However, that is a question for another day. What we now know about conservatively weighted overload and underload baseballs is that they affect how fast an athlete’s arm can move, they affect some end ranges of motion of arm and mid-section kinematics (not the ones you’d think), and they have little to no impact on elbow and shoulder kinematics.
Weighted baseballs do have an effect on throwing kinematics; however, it’s important to note that the magnitude of this effect is very small and likely practically insignificant. For example, as we saw with the point of interest analysis, the greatest mean difference in elbow flexion between any two ball weights was a 2 degree difference between 4 oz and 6 oz throws. The typical range of motion of elbow flexion/extension was around 90 degrees. This means that the biggest effect between different ball weights was about a 2% difference. Doesn’t seem practically significant, does it?
Implications for Weighted Ball Training
Since 3 – 7 oz weighted baseballs don’t change mechanics very much, let’s talk about how they can be valuable. There are a few things I think should be considered: power/speed adaptation; throwing fitness and throwing volume; training efficiency; and athlete comfort.
Many training programs (outside of baseball as well) use underload and overspeed training. The theory is that using underload implements or assisting movements so they can be performed faster would require the body to recruit different motor units and learn to use faster control patterns. If this is the case, then using 3 and 4 oz weighted baseballs would be a low-risk addition to a training program, since we saw that the athletes’ arms moved faster when throwing a lighter ball with relatively no adverse effects on throwing mechanics or stress on the body/arm.
Building throwing fitness and increasing volume in a training program can be done by varying throw frequency, throw intensity, total number of throws, and changing the load. Small increases in ball weight can add more load to a throwing exercise without regressing mechanics, so throwing these leather weighted baseballs can be a good approach to changing total workload (for lack of a more encompassing term) without further increasing total number of throws.
Similarly, since mechanics don’t change with a 7 oz baseball (2 oz overload), maybe it is possible to get similar training effects with fewer throws by using the heavier ball. When I head into the weight room and set up the bench press for a few sets, I can reach max exertion with fewer reps if I use a heavier load. So, if I have less time to do a workout or a warm-up, maybe I would load up the bar a bit more than my typical load if I knew it wasn’t going to adversely affect my movement.
A movement like a throw in which every throw places the arm under a high degree of stress is more complex than the bench press, which supports the use of a light overload even further. If I can place my arm under less high stress repetitions and get a more efficient training effect by using a light overload baseball, that sounds like a good idea.
A training study (or multiple training studies) would be required to be sure of these effects, but knowing that mechanics do not change significantly when throwing 3 – 7 oz weighted baseballs makes this a less risky idea to pursue.
If an athlete really enjoys using overload or underload baseballs to enhance how their body and arm feel for competition or training, we can now say with confidence that their mechanics shouldn’t change as a result (at least not acutely, intra-session). Knowing this, if an athlete enjoys using these different ball weights, a coach or trainer should be comfortable with their use.
We have not observed any significant adverse effects of using 3 – 7 oz leather weighted baseballs for throwing. This is based on two peer-reviewed, controlled studies of pitching mechanics.
If you, an athlete in your training program or team, a teammate, or anybody else is comfortable using leather weighted baseballs for pitching, there is no reason not to (if they’re used responsibly). With proper warm up (i.e. don’t throw a 7 oz with max intent right when you arrive at the field), proper on-ramping (start with conservative volume and increase from there), and proper monitoring (if the overload/underload throws seem to be associated with unusual soreness or pain, adjust volume/intensity appropriately), there is no compelling reason not to incorporate these implements.