DCDS Project ID: DCDSx-MATH-099 SEFMD Project ID: 2263 Category: Mathematics Senior Division https://www.istockphoto.com/phot os/spinning-baseball Mathematical and Statistical Analysis of Spin in Baseball Introduction and Purpose https://rocklandpeakperformance.com/baseba ll-spin-axis-spin-rate-spin-efficiency-explained/ https://www.researchgate.net/figure/Forces-on -a-spinning-baseball-in-flight-The-drag-force-FD-acts-in-the-v-direction-the_fig1_259103270 Note: All photos, graphs, images are created by researcher unless cited below What is Spin? ● As pitch is thrown, the ball spins back pushing air behind in creating a equal magnus force up ● Forces on a Baseball: The drag force acts in the opposite direction of v. The magnus force acts as the cross product of omega, the spin force, and the velocity force. Gravity acts down Types of Spin ● Transverse Spin- Perpendicular to Axis- movement in direction front of ball ● Gyro Spin- parallel to velocity- no movement Purpose ● Modify/Improve understanding of baseball, and the way a baseball spins and travels ● Analyze how spin plays a role in a various different types of pitches ● Improve existing tracking technologies that may be miscalculating spin rates (improve how data is collected) ● Discover new types of movements on the baseball previously unaccounted for ● Help players strategize how to spin the ball to maximize output/success Research Summary ● I collected online data from various MLB websites as well as from university laboratory experiments to conduct many different kinds of mathematical and statistical analyses in order to refine my understanding on how spin works, how data about spin is collected, and how to improve existing theories about types of spin 2 Exploration of Spin Rates ● ● Top Left: Fitted probability function for all swings where Miss (1) or Not-Miss (0), and logit function log(x) = log(x) – log(1 – x) between the variables ○ Color represents fitted probability to miss ○ Can compensate for speed with spin Top Right: Statcast data, plotted average spin rate vs average velocity ○ Too many data points to make proper conclusion, even when broken down by pitch 3 Data Collection and Analysis ● ● Compiled dataset of statcast 2015-16 pitches (n=1,286,980) Calculated the probability of a swinging strike based on spin rate and velocity for each pitch ● Plot has mean spin rate and one standard deviation above and below the mean Analysis ● For all 3 types of fastball, the greater the spin rate, higher probability of swinging strike ● Cutter- faced greatest disparity between high and low spin rate- not effective swing and miss pitch without spin ● Sliders- had more swings and misses as spin increased ● Changeups- Spin rate not as effective- dependent on fastballvelocity/arm angle play bigger role than spin rate on swing probability 4 Pearson residual R = (O – E) / sqrt(E). ● I utilized a generalized additive model (GAM) to predict the probability of a miss for each pitch using the velocity and spin rate data ● For all fastballs, used expected number of misses from the fitted model (E) and actual number of misses (0) ● To compare these values, I computed the the Pearson residual ● Left: Displayed model of residuals for 12 pitchers Analysis: ● Velocity and spin are incomplete descriptions of a pitcher's success ● Large residuals- disparity due to several outside factors due to to pitching motion and pitch location ● I also constructed parallel boxplots of the spin rates from 2018 pitchers who threw more than 1500 fastballs Analysis: ● There is variation in the spin rates within pitchers, but there is also sizeable variation between pitchers 5 Comparing Tracking Technologies to Analyze Spin ● ● ● ● ● ● ● ● Tracked 33 pitches from collegiate level pitcher, and analyzed their metrics and data accuracy Total Spin: Pitch Logic correlation (R=0.936). Rapsodo (R=0.979) Correlation doesn’t examen full differences in instruments Used Bland Altman Plot- to measure differences in means from both. Allows us to see bias or paired differences clearly Mean difference in spin rate (pitchLogic spin minus Rapsodo spin) is 8 rpms ± 30 rpms Spin Efficiency- Measures % of total spin contributing to movement Strong correlation between Rapsodo and Pitch Logic (R=0.985) From Bland Altman Plot- mean difference in spin efficiency3.15 percent ± 2.7 percent (bias is small) Overall, Pitch Logic tracks well under established radar units like in terms of velocity, spin rate, spin efficiency, and spin direction ○ Further studies need to be done whether these measures can compare to other units, and whether it accounts for all type of spin 6 Hawkeye Technology Spin Data to Track Non-Magnus Movements https://www.baseballprospectu s.com/news/article/62912/not-j ust-about-magnus-anymore/ ● ● ● ● ● ● Table reveals improvements in technology, Hawkeye measures trajectory as well as 3-D Spin axis Direction of Magnus determined, and direction of actual movement measured I created a graphic that displays total lift and side (magnus and non magnus, red/blue) movements in xz plane of a sinker and fastball which add up to resultant green vector Large reduction in spin efficiency needed to get same shift in movement as 2S fastball Pitchers have similar spin axes (from armslot) but very different movements Differences between magnus and actual movement give evidence for a new Non Magnus movement previously accounted for (aka Seam Shifted Wake) 7 Alternative Method to Track Non-Magnus Movements ● ● ● ● ● ● I conducted a new analysis directly comparing the direct spin axis to inferred axis (presence of non magnus forces) Created 3 graphs display axis shift All 3 sway to one side or another based on the pitch (changeup, fastball, slider) or left or right handedness Next figure plots the same pitches but normalized to show correlation of a pitch’s total movement with shift in the axis Large shifts in axis are often correlated to less movement, especially for fastballs In flat region of sinker plot from -20 to 20 degrees, non magnus force reduces the magnus force lost from gyrospin ○ Does not occur for 4 seam fastball 8 Data Analysis Comparing Types of Spin with Different Pitches Transverse vs Total Spin ● Key: Blue Tri- FAST, White- SINKER, Light Blue CHANGE, Black Squares- SPLITTER, Black Ast- CUT, Green-SLIDERS, RedCURVEBALLS ● ● ● ● ● ● ● All spin is not alike, and has different components of gyrospin and transverse spin Below I plotted, the transverse spin vs total spin using 2012 MLB data Greater gyrospin would fall below the line Scatterplot considerably random from random measurement error Results: Fastballs/changeups have less gyro spin while sliders/curveballs have it significantly (surprising) I created a histogram of spin differences from fastballs/changeups (blue) and breaking pitches (red). ○ Fitted a dotted curve using a normal distribution with mean=0, and standard deviation of 500rpm Mean of 0 also indicated that points in first figure don't follow the line if useful spin is identical to total spin. ○ Confirms that technique used to relate useful spin is correct (points do follow line if useful spin=total spin) Curve not linear, so doubling transverse spin won’t necessarily double movement 9 Conclusions and Applications ● ● ● ● ● ● ● ● ● Importance of Spin, Spin rate, and Spin Axes in Baseball From the spin rate and swinging strike probability analysis, depending on the pitch type, spin rate can either be very important at inducing swinging strikes, or not play a central role at all. Tracking technologies for measuring velocity, spin rate, spin efficiency, and spin direction, such as Pitch Logic perform well under established radar units but are often unreliable due to various discrepancies Spin rate needs additional context of velocity and spin axis Seam Shifted Wake describes non-magnus movements, (vertical motion by changing orientation of seams relative to spin axis) Laboratory experiments suggest this axis shift which is enhanced with gyro spin Fastballs and Changeups appear to be consistent with their spin being useful, but curveballs and sliders have varying degrees of gyrospin I developed a technique to separate the spin into transverse (or useful) and gyrospin parts. Avenue for Further Research ○ Refine the 3D spin axis to compare whether the reduced movement is really a result of spin efficiency. ○ Analyze the seam orientation and see how SSW depends on it -Only scratched the surface, and there is still a lot that is unknown!!! 10 References 11