Player evaluation and motion capture technology

Yesterday, Brett Hainline wrote about the vast troves of data that motion capture technology will soon bring to analysis of the NBA. The possibilities are both exciting and a little overwhelming. In imagining and understanding those possibilities, one can look to the example of soccer where motion capture technology has existed for more than a decade.

Obviously, the two sports are diametrically opposed in certain ways. Most significantly in terms of the volume of scoring outcomes and scoring chances but also in the number of players having their motion captured. In both respects, the more frequent scoring and the fewer players not in possession of the ball, the differences between the sports (as well as the example of how soccer clubs have made use of the range of data) could work to basketball’s advantage.

TrueHoop at MIT Sloan Sports Analytics Conference

During today's Soccer Analytics panel, Steven Houston, Head of Technical Scouting & Data Analysis for Chelsea, described the difficulty of finding the value of the myriad actions that take place on the pitch. With 11 players running, passing, and tackling for 90 minutes and their actions culminating in relatively few scoring outcomes, Houston had to work back from those outcomes to identify which actions were valuable in leading to their creation.

Basketball doesn’t have that problem. There’s already a common understanding of the value and importance of the actions not just of the player in possession of the ball, but of the positioning and movement of his teammates, as well as the positioning and movement of the opposition. Whereas the culture, both internal and external, surrounding soccer might generally overvalue the importance of ground covered in evaluating a player’s work rate, the basketball culture, at least at the NBA level, appears to have, both internally and externally, a fairly accurate and generally agreed upon sense of the importance of such off-the-ball actions as spacing on offense and providing help on defense. The success of Sebastian Pruiti’s work at NBA Playbook (work complimented by Mark Cuban during the basketball analytics panel) demonstrates this to some degree.

What basketball doesn’t have, yet, is a way to quantify those off-the-ball actions at an individual level. And it may not ever be possible to discretely identify what a player brings to spacing or help defense and what portion of the value in a player’s performance in those areas could more fairly be credited to coaching. Of course, significant changes in player performance in these areas as players change teams and teams change coaches could provide useful information in that regard.

Gavin Fleig, currently Head of Performance Analysis at Manchester City, revealed another potential value from motion capture technology: player-specific measures of aging. When Fleig held the same position at Bolton in 2004, they used the player movement information available to them to determine that midfielder Gary Speed, though almost 35 years old, had not seen any significant decline in his movement. This information overrode the general (and common sense) reluctance to purchase a 34-year-old midfielder and allowed Bolton to identify and acquire an undervalued asset, a player who went on to make 121 appearances and score 14 goals from the club over the next four seasons.

Both the Phoenix Suns and the Detroit Pistons have recently demonstrated the impact a good medical staff can have on a team’s on-court performance, analytics have given us a good understanding of how players (as a group) age, and there are indicators of aging in basketball (Offensive Rebound Rate, Steal Rate, Blocked Shot Rate, Free Throw Rate) that can be derived from box score stats. The ability for teams to move beyond assessing a player’s age in terms of all players and instead assess the player in terms of his personal movement history could provide a significant advance.

This weekend, both Daryl Morey and Kevin Pritchard separately identified the greatest challenge for a General Manager of an NBA team to be understanding that being above average was not sufficient for success in the job. The job of a General Manager is to build the best team. Pritchard specifically discussed the importance of the proximity of a team to a championship when identifying an acceptable level of risk to assume in a transaction.

It’s these issues, the marginal increase in probability of success or a slightly greater certainty of risk assessment, that have dominated the 2011 MIT Sloan Sports Analytics Conference regardless of the sport under discussion. The advances in the field of analytics and the impact of analytics on the practical sports world have taken the discussion deeper than identifying the difference between a good player and a bad player or whether teams should invest in internal analytic work. The future practical gains will largely occur in the margins of existing work, those gains will increasingly be the result of highly technical work with data sets, and, as the work becomes increasingly internal, much of it might be invisible to the outside observer.