Reviewing: Berri D. 1999. Who is ‘Most Valuable’? Measuring the Player’s Production of Wins in the National Basketball Association. Manage. Decis. Econ. 20:411-427.

Logically, there are many people who will need to understand what players are more productive than others in order to determine who should play, what players should be acquired, or what trades should be consummated. In this study, Berri attempts to draw a correlation between a player’s statistics and team wins. One award, the IBM award, essentially is the summation of an individual’s positive statistics less his negative statistics. Unaccounted for in this method s that various statistics carry different values (for example, a steal may be more valuable than a rebound).

Berri makes the argument that a team’s scoring is affected most by how the team acquires the ball, its ball handling efficiency, and the likelihood of converting possessions into points. Give this theory, the following team and player statistics were studied:
points-per-shot
opponents’ points-per-shot
free throw percentage
opponents’ free throw percentage
free throw attempts
offensive rebounds
defensive rebounds
assist-turnover ratio
opponents’ assist-turnover ratio
turnovers
opponents turnovers
field goal attempts
personal fouls

Interesting to note in the above list of statistical categories is that some are team statistics while others are individual players statistics. After mixing these two general groupings, it may be more difficult to determine an individual’s independent contribution to the team.

Another big factor that Berri included was the team’s tempo arguing that statistics will vary based on the tempo. A team playing at a faster clip will likely have more opportunities to increase its box score values. Taking team tempo into account while studying player performance through the above list of statistical categories, Berri decides that assists and personal fouls have limited impact on a player’s contribution to team wins. This was determined by calculating each statistic’s marginal value, which should indicate the impact each has on wins. The six ratios used in the equation to determine marginal values were points-per-shot, free throw percentage, assist-turnover ratio, and then the opponent’s equivalent to the previous three. In the paper, it was unclear how Berri went from these statistics to the marginal impact of three-point field goals made, two-point field goals made, assists, turnovers, free throws made, and free throws missed.

Although most of Berri’s reasoning is logical, some of his results seem counter-intuitive. Most significantly, it is hard to understand how assists and personal fouls can have little impact on an individual’s contribution to a team. First, out of the last three MVP’s in the NBA from the 2004-2005 to the 2006-2007 season, an assist expert, Steve Nash, won the MVP two times taking his team to higher levels in the playoffs than in the previous seasons. Also, an assist is only awarded after a successful basket; therefore assists should correlate highly with points. Berri does not seem to consider causation; for example, in the case of Nash, the fear of the pass may force defenders to play him differently, allowing him an increased opportunity for a jumper or a drive. Second, a player’s personal fouls can dictate how much playing time he gets. For example, receiving two personal fouls in the first quarter generally forces the coach to bench that individual until the second quarter. The case is similar when a player receives 3 personal fouls in the first half. Given these circumstances, a lower personal foul total would have a positive effect on quantity of playing time.

I will use this space (crossposted on www.babyhook.com and numbers.babyhook.com) to share my opinions on topics from sports analytics. Most of what I talk about will surround basketball; however, it will not be limited to purely that one sport.

There are many questions in sports–especially the hard questions—of which I feel we can come closer to answers by logically analyzing the numbers and previous statistics. I hope to address the following plus more: a player’s worth, issues of competitive balance, understanding specific stats, star power, coach and ref biases, and fan power.