Player Profiling

The following description of my job duties and responsibilities related to Player Profiling efforts was taken from the golf sports science research section:

Player Profiling – principal contributor to the characterization of players based upon swing, impact conditions, and resultant golf ball trajectories; perform data reduction and statistical analysis of test results and conduct comparisons with simulation results to characterize golfers based on physical measurements, player, club, and ball kinematics, and/or kinetics of the player and club; lead the development of future golfer characterization measurement systems based on accurate interpretation and communication of analytical and experimental data.  

The golf sports science research section provides a lot more details on the history of measurement systems used and also details why and what changes were made resulting in more efficient player profiling efforts.  This section will demonstrate some of the analysis techniques used in player characterization efforts.  All information provided here is taken from presentations at conferences or technical publications, as that data is public domain and not subject to non-disclosure.

Following is a summary of the measurement systems that were used to address the first of the job duties and responsibilities for Player Profiling projects – principal contributor to the characterization of players based upon swing, impact conditions, and resultant golf ball trajectories.  


Initial player characterization studies revolved around OBD II data.  As a reminder, there were 12 different measured strains that were recorded on-board the club and downloaded to a computer after a series of swings.  Six independent forces and moments were calculated from each group of rosettes at both the tip and butt ends of the shaft:

  1. 3 independent forces
    1. Px – Axial Force
    2. Vy, Vz – Transverse Shear Forces
  2. 3 independent moments
    1. Tx – Torsion
    2. My, Mz – Bending Moments

An example of the loadings of two different swing types with similar peak magnitude values for Mz is given below.  While the peak values of Mz are very similar in shape and magnitude during the early downswing, they are very different at impact.  Also, the My moments are not nearly as similar throughout the swing.  This just goes to show the difficulty in trying to characterize swings by analyzing only a portion of the dynamic swing for a variable like peak Mz moments.


The following graph is another example showing 3 different types of swings.  These are plots of the measured microstrain used to calculate the Mz moments.  Very different types of swings both in peak magnitudes, shapes of dynamic swing, and relative timing of events in the swing.


The following graph is a comparison of a high handicap golfer vs. a low handicap golfer. This shows the average and confidence intervals for 5 different swings between the two different types of golfers.  The lower handicap golfer exhibits shows less variability throughout the dynamics swing, as most would probably guess.  However, it also shows that the variability is more during the backswing, transition, and early downswing. During the latter phases of the downswing, when centripetal acceleration and inertia have overtaken the player feedback control system, the variability is not nearly as large.


The previous graphs showed how the OBD results varied in a number of different ways: peak magnitudes, relative timings, shapes of the dynamic values, values at impact, etc. This was true for all 12 variables that could be analyzed.  The My and Mz bending moments were two of the variables that showed the most promise for characterizing golfers. Another meaningful analytical tool was to plot the variables against one another during the dynamic swing.  The shapes for these analyses provided a powerful pattern analysis technique for characterizing swings.  This is an example of the same player swinging two different clubs with variation in the shaft stiffness.


While all of the previous plots were taken from OBD II clubs, there was also synergistic work with motion capture data analysis.  One of the main reasons for including player kinematic analysis for characterization efforts is that the player is the engine driving the club.  They are providing the actual input to the club kinematics.  As the following chart shows, player profiling efforts have to include player kinematics in order to understand the resultant dependent outputs of OBD and CPAS technologies.playerdependencyKinematic analysis can provide many different types of variables to analyze – any 3D position, velocity, and acceleration for any segment and/or club, or any 3D rotation, angular velocity, and angular acceleration for any segment and/or club.  One of the more commonly used kinematics output for analysis is kinematic sequencing.

bottomFollowing are two different kinematic sequencing plots for two different golfers: a professional golfer on the top and a low handicap golfer on the bottome.

playerkinseqplayerkinseqamWhile quick glance shows that there are similarities in the two graphs, closer inspection reveals a lot of differences.  The following graph shows a more detailed break down of the professional golfer’s kinematic sequence profile.  This plot shows that a professional golfer has sharper slopes in their angular velocity curves, which indicates higher acceleration profiles.  They also demonstrate higher peak magnitudes for each segment as well.  The timing of the peaks of each segment are also more optimal, with each distal segment peaking later than the more proximal segment.


One important result of proper kinematic sequencing is the effect that it has on the resultant club head speed at impact.  The following graph shows the measured clubhead speeds for the swings of the same two golfers presented above.  We can see that despite similar slopes of the club head velocity during the downswing, the professional golfer achieves an approximately 20% higher swing speed due to the more optimal and efficient kinematic sequencing profile.


The following graph is an example of a plot of the club head kinematics only during the dynamic swing.


The golf sports science research post detailed why their was a transition from OBD II strain-gage shaft measurements to an OBD custom inertial sensor (CIS) data acquisition system.  The details of what was changed and why can be found in the OBD II CIS post.

The following graph shows a good example of why the transition was made.  This plot shows the differences in the left forearm to shaft angle throughout the swing for the same two players shown above.  The previous plots showed differences in kinematic sequencing profiles as well as club speed throughout the swing.  This plot provides more powerful data as this is the last player input powering the club motion.  We can see some major differences between the two players: a much larger magnitude difference just before downswing, including an increase in the angle in the professional just as downswing is initiated, and the release rate differences shown by the slope of the line during the downswing.  This provides evidence that an angular velocity measurement of the release rate in the club with an OBD II CIS would be very powerful for player profiling…which it is! 

The following graph shows direct OBD II CIS measurement of the release rate as well as the closure rate up to impact.  This system allowed us a very powerful tool for player profiling as it could be used in any club and in location to gather data on all types of golfers.  The system is completely transparent to the player unless they were told they were swinging in.  The OBD II CIS system was the missing piece that linked all 3 of the Product Player Matching projects: Player Profiling, Digital Human Modeling, and Club Fitting. rotations2

Multi-sensor swing studies (MS3) with both optical motion capture and the OBD II CIS system verified that both systems were providing the same results.  This allowed me to link our player profiling studies with our dynamic swing modeling efforts, where we used optical motion capture data outputs as high fidelity inputs for forward dynamic simulations in a virtual product development capacity.  These digital human swing models provided powerful design tools to help with the development of analytical and experimental tools for the fitting of golf clubs based on player profiles.

Following are highlights from my player profiling efforts:

  • Over 5 years of experience in golfer profiling studies, including responsibility for the design, coordination, and compilation of golfer-club interaction testing. Designed an advanced testing protocol for collecting, extracting, processing, synchronizing, and analyzing player and club kinematic data from multi-sensor swing studies (MS3).
  • Led exploration of golfers’ sensitivity to key club design attributes based on a thorough understanding of player swing characterizations through kinematic and kinetic analysis of the golf swing.
  • Utlilized pattern classification techniques for characterizing golfer swing profiles based on physical characteristics, player kinematic and kinetic parameters, and shaft/club 3D motions, angular velocities, and accelerations.
  • Conducted quantification of feel studies with a custom grip pressure design to investigate the golfer’s feedback function using the elastic and inertial properties of the club throughout the swing, sound characteristics at impact, and the shaft’s modal characteristics through impact.
  • Over 3 years of experience using Analytical Hierarch Process (AHP) techniques for more efficient player performance testing by eliciting and quantifying subjective judgments and analyzing pairwise comparisions in player-club interaction tests.

Leave a Reply

Your email address will not be published. Required fields are marked *