With the cancellation of the minor league season and being left off of the 60-man roster, I have been in search of ways to simulate in-season pitch development as well as quantifying that progress in addition to the live AB’s that I throw bi-weekly at R&D. Unfortunately (or fortunately), I have a ton of data points in my head to strive for that are significant outliers and amongst the best pitches in baseball, but I have realized that due to release height/side, velocity, throwing posture, etc. some of these pitch fx numbers I am chasing are unattainable. Furthermore, my data intern, Cam (@k_camden) and I have been analyzing inflection points in terms of horizontal & vertical movement as well as velocity (extremely important) to see where I may recieve my biggest return on investment. As well as applying this knowledge to my trainees in their own respective opportunities for growth during these developmental windows.
Similar to our MLB pitcher arsenal/pitch comp work we did (Here) we are utilizing all of Major League data from 2019 as our source of reference. We decided to plot wOBA as well as Whiff rate, as these are extremely strong determinants of stuff as it pertains to optimal outcomes of swings and misses and weak contact. At first glance we graphed in isolation, but then decided to shift to normalized rates (36% whiff for sliders and a corresponding .286 wOBA). For example, instead of graphing wOBA 0 to X we would then use .286 as the median line for sliders and present data as either negative or positive from average.
From here, it became evident to myself that heat map visuals as it pertained to horizontal and vertical movement would be very beneficial, which @k_camden was able to integrate.
First and foremost I think it is critical to state and appreciate the complexity behind pitch design. I, myself, have been guilty of oversimplifying the pitch design process especially as it pertains to data points familiar within the industry such as hitter expectation fastballs, deviating a CH 30+ minutes hand-side for increased whiff rate, merging big arsenals with gyro sliders, spin mirroring etc. and have often viewed these as absolutes. However, after diving into this data I realized I need to place my guys (and myself) in position to attack the lowest hanging fruit in terms of where a small change could pay off substantially. Outside of identifying the solution closest to a current data point (velo, horizontal/vertical movement) we are also able to bracket guys in realistic movement categories and not try to reinvent the wheel. As an example pitcher A has a slider that sits around gyro and is in the low 80's:
Now, we can utilize the heatmap and velocity graphs to search for any inflection points around sliders:
Now we see how valuable a 0,0 slider is (not using that as an absolute though)
There is now an opportunity to utilize an inflection point! If pitcher A were to increase his slider to 84+ on average (now about 83) he will be in a situation where movement and velocity predict a ton of success with this pitch, as we see Whiffs greatly increase around this velocity.
Also to be clear, a lot of these inflection points are small, but within the data sets we collected (velo, horizontal/vertical movement) we are certainly able to identify huge margins of return with various pitches, for instance with velocity and curveballs. Again, throwing a curveball 80+ is an incredible attribute and extremely hard to master but just to illustrate the value of big inflection points here is that graph:
How absurd is the value of going from <80 to >82! wOBA absolutely plummets while Whiffs soar. Look no further than Tyler Glasnow as example A of the impact of velocity on curveballs as he averaged about 84 mph last year, and despite only having the 120th "most" curveball depth (factoring in gravity) he accumulates a ridiculous 43.9% whiff rate.
Credit to Made the Cut Youtube:
An area where I think inflections can really help is in fastball categorization. As I have discussed before I am not a fan of Rapsodo data as individuals will cheat to get to certain axis requirements simply because the built in algorithm will only give certain pitches an "outlier" movement profile because it is flawed. Moreover, being able to take true Trackman data and find where guys can benefit from added or subtracted horizontal/vertical is advantageous so guys are not cheating to differing postures (Prior Blog) to obtain certain movement characteristics.
We can see here despite having an "average" fastball as long as we do not decrease carry below 15 and can add some run or cut the opportunity for success greatly increases. Thankfully, with seam manipulation, or very slight axis changes we can create a bit more cut or run without reinventing the wheel. Again, this process is complex especially as it pertains to gaining velocity which is why our process is so thorough and never viewed as a linear process, but you cannot ignore the benefits velocity gives to every pitch type.
As Cam and I continue to dissect my data points, we have been able to iron out a few key points where I can return substantial value with small changes. As touched upon earlier, at this point in my career I cannot reinvent the wheel, but I do feel confident in subtle adjustments as it pertains to horizontal/vertical movement as well as velocity.
I will first start with my sinker, which is noted as a heavy groundball pitch with a limited number of whiffs; furthermore, Cam and I set out to figure out where my best margin for return would be amongst those categories. After analyzing the horizontal/vertical heatmap, we realized we were in a good position:
From here, we needed to factor in velocity:
With my average velocity sitting around 89.8, we do see a decrease in wOBA past 91, with whiffs climbing as well, so we made a sound commitment to velocity training while being sure to retain the same movement profile. While the velocity is not eye popping it is certainly an improvement for me especially in a training setting.
With my slider, we have seen progress in horizontal break, but are striving for even more side spin influence by limiting backspin. This is a pitch that I have previously compared to Chaz Roe frequently, but as noted factoring in release points, posture, etc. I needed to individually assess this goal to be elite in comparison to other low posture throwers:
When we ran this data in terms of velocity there were no massive inflections evident, but we certainly lose value when velocity goes below the upper 70's range. So being able to maintain this movement while keeping velocity above 80 mph is beneficial as hitter expectation with additional movement relative to a velocity decrease should not be in play.
Lastly, we looked at my changeup, which brought to light the fact that I should utilize this pitch more to both handed hitters. With this, I know that the horizontal number will not exceed my sinker, and being able to obtain depth would require me to topspin the ball a bit, but with the help of a few people, notably Brent Honeywell, I have been able to do it fairly consistently.
Thank you guys very much for reading! I want to recognize Steve Nagy and Camden Kay as these two push me daily in my thought processes as both a player and trainer of other professionals - they both are posting great blog content as well. Please feel free to reach out for further discussion either by email: firstname.lastname@example.org or twitter: T_Zombro24