The Relationship Between Ball Weight and Ball Flight Characteristics

by Jordie Henry, with clutch help from Ilan Joffe

Introduction

If you are anyone in baseball, you know that much has been observed about ball flight tracking over the last several years in the baseball world. The rise of technology like TrackMan, Hawkeye, and Rapsodo has changed the way pitchers develop for the foreseeable future. Such aspects of tracking ball flight include in-season and live game pitch analysis, training and development-centered pitch design processes, and areas where those concepts might overlap as an athlete monitors their personal progress over time. 

The types of questions, analyses, and conclusions that can be drawn from this tech are seemingly endless. For specific examples of how these tools can be utilized, check out previous articles from the blog here, here, and here.

In this post, we will look at yet another and possibly stupid example of how this tech and its data can be used.

*For sake of specificity and to avoid going off on tangents, the rest of this post will carry on under the assumption that the reader has a general understanding of basic pitch metrics. 

So,

Now that all the casuals have exited this blog post, let's get to what this post is actually centered on. 

As mentioned, there is a tremendous amount of research that has been done on ball flight characteristics at all levels of baseball from published journal articles to nobodies on the internet like myself playing with a Rapsodo, collecting data, and maybe (but maybe not) making something of it via blogs that a handful of people read. No matter the level of professionalism, a lot of the research is pretty cool. Actually, most of it is pretty cool.

Most ball flight analyses focus on niche topics that look at pitch characteristics as they are paired with varied intent levels, types of grips, types of seams or seam orientations, ball types, pitch velocities, spin directions, foreign substance application, external environments, changes in a pitcher's delivery or mechanics, variabilities over the course of a season... you name it. I'm sure I am missing several types of studies that I have browsed and enjoyed, but you get the idea. There are a lot of directions you can go when a ball is thrown 60 feet 6 inches and you have an expensive machine to track its path.

However, I personally have not yet seen anything looking at the relationship between pitch characteristics and varying the weight of the baseball being thrown. In simple terms: what happens when you fire up the Rapsodo and throw weighted balls off the mound in front of it?

Only one way to find out.

The Experiment

To test this question, I fired up our team's Rapsodo unit at Scotland Campus and had our guys throw weighted balls off the mound in front of it. Luckily, our test subjects (the players) were already participating in a velo program with designated weighted ball velocity training days, so it was easy to just let the Rapsodo run and collect the data for later use.

This specific velocity training group consisted of 7 pitchers all between the ages of 18-20 years old. Each pitcher had one "Weighted Ball Mound Velocity" day per week for four weeks. (side note: COVID-19 happened and cut this velocity training phase in half.)  During the "Weighted Ball Mound Velocity" days, each pitcher threw sets of 3-5 max effort fastballs off the mound with leather baseballs ranging in weight from 3 ounces to 7 ounces. The first throw of each set was performed at 75-80% effort to give the pitchers time to become acclimated to the new ball weight between each set. The format of the weighted ball mound portion of these training days looked something like this:

5 oz baseball: 3-5 throws at max effort, first throw at 75% effort

6 oz weighted baseball: 3-5 throws at max effort, first throw at 75% effort

7 oz weighted baseball: 3-5 throws at max effort, first throw at 75% effort

5 oz baseball: 3-5 throws at max effort, first throw at 75% effort

4 oz weighted baseball: 3-5 throws at max effort, first throw at 75% effort

3 oz weighted baseball: 3-5 throws at max effort, first throw at 75% effort

*during Week 1 of this velocity phase, the spread of weighted balls only ranged from 4-6 oz. After Week 1, each pitcher performed a full spread of weighted balls from 3-7 oz. 

While these pitchers performed their throws, Rapsodo ran live as I changed player profiles between sets to tag each individual pitcher with their respective throws. Unfortunately, Rapsodo does not have the option to tag ball weights on pitches, so I had to manually tag the ball weight to each throw with pen and paper later on via Google Sheets.

The result was four weeks worth of raw data comprised of 506 individual pitches, each tagged by player and weight of the implement. In addition, each low intent throw at the beginning of each set was gathered, tagged by weight, and designated as "low intent" in the dataset to examine the possibility of a difference among intent level and weight of the implement thrown.

Limitations

The set of baseballs used consisted of leather weighted baseballs from Driveline Baseball for the 3, 4, 6, and 7 oz balls. For the regular 5 oz. baseball, we used our standard game ball: a Wilson A1010 Pro.

As you can see, each weighted ball is accented with its own color, most noticeably on the seams. Would this cause issues with Rapsodo's ability to read all ball weights? This is our first real potential limitation. If one of the weighted balls was unable to be tracked and have data gathered during its flight, that would render an unfortunately large amount of data useless for this study. (Spoiler: the 6 oz. ball and its orange nature had difficulty gathering as much data as its 3, 4, and 7 oz. teammates.)

The second limitation was the difference in ball type with the 5 oz. baseball. All of the Driveline weighted balls are made of the same leather, have the same seam height, and are the same circumference. With our Wilson game ball, the leather is tackier (better grip), and the seams are noticeably higher when wrapping your fingers around the ball. 

Third, some test subjects included in the velo training did not participate in Week 4 as they went home when COVID-19 ramifications came to a head. For these players, there is a large amount of data that we weren't able to gather, specifically with the 3 and 7 oz. balls as those ball weights were not used by any pitcher during Week 1, either.

The last limitation was a general disregard for command during the velocity phase. In this phase, a target is set behind the plate, but we are not concerned with throwing strikes at this time. We are concerned with throwing hard. As an effect, there were a decent number of throws that were unable to be tracked by Rapsodo because they were thrown too far outside of it's scope of vision. Most throws that "didn't read" only garnered velocity readings and no other metrics. A small handful didn't even read velocity.

A last note on the limitations is that any blatant misreadings were tossed out. For example, spin rates of 2800+ RPM or 600 RPM in a pool of otherwise normal spin rates around, say, 2000 RPM were discarded. There were roughly 10 of these pitches that were not considered in the dataset.

Hypotheses

My initial hypotheses targeted a few different pitch metrics. 

First, I guessed that as ball weight increased, spin rate would decrease. The logic behind this was based largely on feel of using those balls myself. Throwing an underload (3-4 oz.) ball just feels easier to spin at release. They also seem to carry more in catch play and resist gravity more than a baseball. On the other end of the spectrum, it was hard for me to conceptualize that spinning an overload (heavier than 5 oz.) ball is as easy as a regular baseball. Imparting the same or less force behind an overload ball when playing catch seems as if it would decrease overall spin, even relative to velo, due to its weight and literally not being able to pull down on the seams as much. This especially made sense when thinking about playing catch with a ball as heavy as an 11 oz. It just doesn't feel like it spins the same out of the hand. 

Something to be aware of and take into account is that Bauer Units come into play here. If you are not familiar, Bauer Units normalize spin rate relative to velocity. As velocity increases, spin rate generally increases. That being said, players would obviously throw a 7 oz. ball slower than a regular baseball, so it would probably spin less. In this context, I will be tracking change in spin rate as it relates to velocity using Bauer Units.

Example of Bauer Units: 

2200 RPM/92 MPH = 24 BUs 

... but ... 

2200RPM/85MPH = 26 BUs

My second hypothesis was that the overload balls would have a larger gyrospin component. In layman's terms, the heavier balls would be "cut" more. In metric terminology, the spin efficiency would be lower. This was also due not only to my own feel and experience but also due to the tendency of players to get on the side and "cut" the heavier balls during plyocare work, especially with the 450g blue ball and 1000g green ball during arm action-focused drills. 

Lastly, I hypothesized that there might be some difference in spin direction among varied ball weights. I wasn't sure what direction change might be associated with different ball weights, but I expected to see some vertical or horizontal tilt in spin direction as the ball transitioned from heavier to lighter implements.

Other than those three ideas, I had no other expectations or guesses as to what might happen with the experiment.

Results and Analysis

The table below includes all seven players that participated in the velocity training phase. It includes high intent throws only, and does not include the first throw of each set at 75% effort. 

Group Average Pitch Metrics

*Release metrics are omitted. Averages from release included a 6.0 ft release height on every ball, as well as a release from the side at 1.8 ft. Also, release metrics, by label, do not play as much of a role in gross ball flight metrics as the ones shown. There were also no glaring differences in the other release metrics that were left out.*

The first thing to note is that the velocity spread among the ball weight spread is consistent. As a general rule of thumb, as ball weight increases/decreases, velocity may increase/decrease with it by roughly 3 mph. This is exactly the case except for the difference in 4 and 5 oz. balls at an average of 2 mph. Different, but not really.

The most obvious variable here, though, is the spin rate discrepancy among the 5 oz. ball and the rest of the ball weights. Not only does the 5 oz. raw spin rate obliterate the raw spin averages of the other balls, the Bauer Units are significantly higher. For context, MLB average spin is 24 BUs. As an entire group, spin rate relative to velocity for the 5 oz. ball was actually above MLB average by roughly one Bauer Unit. This is odd considering seven different players are included in the dataset. Even more confusing is the far below average Bauer Units tallied on the other balls at 20 and 21 BUs for the underload and overload balls respectively. This is very low spin that typifies pitches like sinkers and changeups.

Relative to spin efficiency and the gyrospin component, the only significant difference is among the overload balls at 5% less spin efficiency than the 5 and 3 oz. balls, and 3% less than the 4 oz ball. This is especially significant when taking into account that the relationship between the spin efficiency metric and gyro degree is not linear, even though the two metrics mean mostly the same thing. In essence, the higher a pitch gets in terms of spin efficiency, the more true spin it actually needs per percentage point. Not that it's super important here, but just know that 5 percentage points in this context is quite a lot in terms of overall backspin imparted on the ball. To sum up this bullet point, the overload balls are simply cut quite a bit more than the 3-5 oz. balls due to a larger amount of gyrospin imparted on the ball at release.

In terms of spin direction and tilt, there is really no difference among ball weights. This is actually surprising, as it seems the amount of gross proprioception needed to adapt to such wide variances in ball weight in such a short window of time as the player progresses from set to set would need to be substantial, but maybe this isn't so. From most vertical spin direction at 188 degrees to most sideways at 193 degrees, the spin direction never varies more than five degrees. For those not familiar with spin direction in terms of degrees, imagine 180 degrees as 12:00 on a Rapsodo. The range of average spin direction among all balls in this dataset is roughly 12:15 to 12:25, or a range of 10 "minutes." Not much.

To sum up these group averages, we find that velocity spread is normal, the 5 oz. ball is spinning A LOT more than every other ball, overload balls exhibit quite a bit more gyrospin or "cut," and there seems to be no real change in spin direction among any of the balls.

However, anyone familiar with analyzing data knows that group averages or averages representative of an entire sample are not always indicative of what happens on an individual level. 

While these group averages may give us general insight on what happens to ball flight as ball weight changes, they may not tell the story of what happens from player to player in the sample. 

Let's look at an example from Pitcher 1 in the dataset:

Pitcher 1 Average Pitch Metrics

With low and high intent averages included here, it is increasingly easier to see the giant discrepancy in spin with the 5 oz. baseball in relation to the under/overload balls. Pitcher 1 has BUs that are even less than the group averages when it comes to the weighted balls at 18-19 BUs, but are exactly group average on the regular baseball at 25 BUs. 

Something of note here that was not covered in the group examination is the difference in total movement that each ball exhibits. All balls for Pitcher 1 spin at roughly 150 degrees, but the massive increase in spin on the 5 oz. induces about 4 more inches of total movement, mostly coming from a larger vertical break component, or "carry."

Also of note is an asymmetrical velocity spread, which has less to do with pitch metrics but points more to the idea that pitch characteristics are a little more volatile within each player than group averages may indicate. The spread here, from 3-7 oz. respectively, is a decrease of 2.6, 1.4, 3.3, and 1.4 mph as ball weight increases. 

Let's look at an example of player-to-player volatility further towards the other end of the spectrum, where metrics and characteristics have very little, or at least less, consistency with group averages:

Pitcher 5 Average Pitch Metrics

First, let's take note of what is similar to the group average. Here, there is a little more consistency in the velocity spread than Pitcher 1 as it more closely resembles the 3 mph change per ounce. 

Notice also the dramatic increase in raw spin and Bauer Units on the 5 oz. again. As we look at group average and its relationship to player averages, this 5 oz. trait seems to remain constant. This may lead us to theorize that the type of ball itself may play a role in spin rate discrepancy. Remember, the 5 oz. baseball is a Wilson 1010 Pro ball while the other balls are Driveline Baseball leather weighted baseballs. 

Aside from these two characteristics, not much else stays consistent with the group average.

The first deviation from group averages is, ironically, also related to spin rates across ball weights. Recall that in the group average, Bauer Units among the 3, 4, 6, and 7 oz. balls were all 20-21 BUs. Here, when omitting the 5 oz. ball from the data, there is an increase of 1 BU per ounce starting from 21 BUs on the 3 oz. and ending at 25 BUs on the 7 oz. (looking at high intent throws only). Besides velocity in the group average, this is the first linear relationship we've examined between two metrics.

The second deviation from group averages in Pitcher 5's data is the volatility in spin efficiency, or amount of "true spin," as ball weight changes. 

Looking at high intent throws, we can see that:

- 3 oz. spin eff = 98.6

- 4 oz. spin eff = 93.7, a decrease of 4.9%

- 5 oz. spin eff = 96.0, an increase of 2.3%

- 6 oz. spin eff = 95.0, a decrease of 1.0%

- 7 oz. spin eff = 88.7, a decrease of 6.3%

Which means ... 

... Not only within the player, but also in relation to the group. Within Pitcher 5, the variance in gyro degree changes inconsistently. Similarly to the group, the overload balls have less spin efficiency/more gyrospin than the other balls, but actually to much different degrees (no pun intended). 

In the group averages, the 5 oz. ball tallied a 95% spin efficiency while both overload balls tallied 90% spin efficiency. Also recall that the relationship between spin efficiency and gyro degree is not linear. Here, with Pitcher 5, the spin efficiency metric at 96% and 95% with the 5 oz. and 4 oz. balls respectively could be considered similar, while the very low spin efficiency of the 7 oz. at 88.7% has a gyro component that is getting close to the territory of a cut fastball pitch type. 

Essentially, the relationship between overload balls and the 5 oz. ball in the group average - and the relationship between overload balls and the 5 oz. ball with Pitcher 5 - are not consistent, though they could be considered generally similar.

Lastly and briefly, with Pitcher 5, we see a larger range of spin direction, from 204 to 212 degrees, (aided in part by the 3 oz. ball's direction in relation to the rest of the balls) than the group average. Maybe more indicative of changes from player to player is a far more sideways spin direction. Here, Pitcher 5 averages a spin direction of 206 degrees across all balls while the group average was roughly 190 degrees. 

The most important lesson from Pitcher 5's data is that while some of the general aspects of this athlete's data resemble the group averages, they vary to greater or lesser degrees depending on the metric.

Our third and final example examines Pitcher 7, who exemplifies consistent results within his own throws, but again, isn't exactly adhering to group averages that much. In fact, Pitcher 7 strayed farthest from his teammates' averages:

Pitcher 7 Average Pitch Metrics

The most glaring trait with Pitcher 7's data is the large velocity spread as ball weight changes. There is about 5 mph difference between 3 and 4 oz. balls, about 1.5 mph difference from the 4 to 5 oz., and then more than 5 mph between the rest of the ball weights, a far cry from the 3 mph average spread that is typical from ball to ball as the group average exemplified.

Additionally for Pitcher 5, the difference in Bauer Units and spin rate between the 5 oz. and all other balls was anywhere from a 7-10(!!!) BU decrease, while the group average was only 4-5 BUs less than the 5 oz. ball. At that much of a decrease in spin, you are looking at drastically different pitch characteristics and movement profiles. This can most easily be seen in the differences of horizontal and vertical movement between balls in the table above, with a total movement decrease of about 6 inches from the 5 oz. to the 6 oz. ball. 

The last point of note in Pitcher 7's data is his healthy amount of spin efficiency on all ball weights. There is very little gyrospin imparted on any ball, as Pitcher 7 actually increases the amount of true spin as ball weight increases. This is almost opposite of the group average data, as overload balls exemplified significantly more gyrospin than any other ball weight. Here, with regard to the relationship between spin efficiency and ball weight, Pitcher 7 is the outlier. 

Conclusion

What did we learn? Well, the pitch metrics within individual player data varied to different degrees depending on the pitch metric. For example, Pitcher 5 gradually increased his BUs as ball weight increased, and Pitcher 7 gradually increased his spin efficiency as ball weight increased. Neither of these traits were evident in the group averages.

However, there were some consistent results...

Unfortunately, none of these consistencies can reasonably be attributed to ball weight on its own.

The most evident example of this is the blatant spin rate increase with the 5 oz. ball across the entire dataset. An average increase of 4-5 Bauer Units with the standard 5 oz. baseball is something that cannot be ignored. This trait, paired with the fact that the 5 oz. baseball is the one ball in the set that is manufactured differently from the rest of the weighted balls is a limitation that must be considered in the results. 

As mentioned in the limitations above, the 5 oz. ball is made of generally tackier leather and has higher seams. There are a few studies that examine the relationship of seam height and ball type on pitch characteristics, but without going into detail, it is almost obvious that this physical limitation played at least some role in skewing data, especially with the 5 oz. baseball. 

Or, what could be the case is that the 5 oz. baseball on its own registers mostly normal pitch characteristics, while the Driveline Baseball leather weighted baseballs register abnormal pitch characteristics that may include generally below average spin rates. This can be rationalized by the fact that the Driveline balls are generally "slick" and don't have the same grip or feel as any type of regular 5 oz. game or practice baseball. This could be due to the durability that has to be required of these training baseballs. 

Regardless, the material that the over/underload balls are made of are different than a normal baseball, and I don't believe it is a coincidence that the spin rates vary so much among the two ball types. This is a part of the experiment that severely limits or skews data, and points more towards the possibility that ball type might play more of a role in pitch characteristic variation than ball weight.

Even though the data may be skewed to an extent, the initial hypotheses can still be tested.

Hypothesis 1: As ball weight increases, spin rate will decrease. 

This snapshot of the group averages shows that there is no real relationship between ball weight and spin rate across the entire sample. When omitting the 5 oz. ball for reasons discussed above, the only conclusion that can be drawn is that spin rate actually increases slightly, as the overload balls express BUs of 21, while the underload balls spin at a rate of 20 BUs. Still, a variance of 1 Bauer Unit cannot be considered significant, especially across such a small sample of data.

Hypothesis 2: Overload balls will have more gyrospin.

Word this however you want, but the idea is that the overload balls would have lower spin efficiency, a larger gyro degree, or as I said, "more gyrospin."

This could be considered generally true among the group, as 90% spin efficiency is much lower than the rest of the ball weights. Again, this is taking into consideration the relationship between gyro degree and spin efficiency. 

However, we saw with a couple of the individual pitcher examples such as Pitchers 5 and 7 that this is not inherently true. Pitcher 5 showed no significant relationship while Pitcher 7 actually increased in spin efficiency as ball weight increased. 

Hypothesis 3: There will be a difference in spin direction as ball weight changes.

In summary, there was no real difference. Actually, across the entire group and within each individual player's data, there was no significant difference in spin direction aside from two pitchers that exhibited slightly more vertical tilt with the 7 oz. ball. Overall, though, spin direction varied no more than typical fastballs would vary from pitch to pitch in bullpens or game settings. 

So, we found out nothing.

Cool, cool. Does this mean we can take nothing from this experiment? Not really. 

A surprising aspect of this that ended up being lethal to finding real conclusions was the difference in ball type. It is obvious that the Wilson ball and the Driveline balls produce vastly different amounts of spin than we ever thought to consider. So, as briefly mentioned before, is ball type as it relates to changes in pitch characteristics a question that demands more experimentation?

I recently stumbled across a few articles from Barton Smith at Utah State University that examines this. An interesting one that is worth the read can be found here.

Something else that can be learned is that maybe it's not worth going down rabbit holes with weighted balls for purposes of pitch design.

That's right: pitch design. Which was the origin of this experiment.

I wanted to see if weighted implements could help with the "feel" or proprioception of manipulating certain pitch types. 

For example, say the hypothesis of overload balls inducing more gyrospin was worth looking into further. Could we use a 6 or 7 oz. ball to work on the feel of supination or getting on the side of the ball when trying to fine tune a slider or cutter? 

Say the underload balls induced higher spin rates or helped improve spin efficiency? Could we use those balls to help our fastballs carry more? Or help us pronate through a changeup?

This idea came from the concept of using throwing implements that vary in size, such as command training baseballs that vary in circumference (either slightly smaller or larger to work on the above pitch design goals), tennis balls, footballs, etc. 

The answers to those questions won't be found in this admittedly small case study. I've found this to often be the answer: that there is no answer yet. But, in the end, we examined something that I had never seen studied before.

I'm sure some type of study like this exists somewhere. So, if the 6.9 people that have read this far have happened to stumble upon something similar, feel free to send it my way!