Science News
from research organizations

# Basketball as a statistical network

Date:
December 11, 2012
Source:
Arizona State University College of Liberal Arts and Sciences
Summary:
What started out as a project to teach undergraduate students about network analysis, turned into an in-depth study of whether it was possible to analyze a National Basketball Association (NBA) basketball team’s strategic interactions as a network. Researchers discovered it is possible to quantify both a team’s cohesion and communication structure.
Share:
FULL STORY

This is a weighted graph of ball transitions across two games for the (a) Bulls, (b) Cavaliers, (c) Celtics and (d) Lakers. Red edges represent transition probabilities. Player nodes are sorted by decreasing degrees clockwise from the left. The thicker the red line, the more frequently that pass occurs during play.
Credit: Photo by: School of Life Sciences

What started out as a project to teach undergraduate students about network analysis, turned into an in-depth study of whether it was possible to analyze a National Basketball Association (NBA) basketball team's strategic interactions as a network. Arizona State University researchers discovered it is possible to quantify both a team's cohesion and communication structure.

The researchers' findings appeared in an online issue of PLOS ONE.

Jennifer Fewell, a professor in ASU's School of Life Sciences in the College of Liberal Arts and Sciences, and lead investigator on the project, explains that because teams are an integral part of both human and animal societies -- understanding how a team's interactions as a whole affect its success or failure is important.

"We were able to come up with a hypothesis about strategy and then apply network analysis to that," says Fewell. "Often, people simply create networks and then conduct descriptive analysis of them, but they don't actually explain why they would expect an individual in a group to communicate the way they do. We take a different approach by suggesting that there are potentially successful ways to organize your team if you use this strategy then we should expect this network metric to show up as an indicator -- sort of a proof of concept."

The researchers measured two offensive strategies to learn whether differences in offensive strategy could be determined by network properties. First, they looked at whether teams moved the ball to their shooting specialists -- measured as "uphill/downhill flux," and second, whether they passed the ball in an unpredictable way -- measured as team entropy. They analyzed games from the first round of playoffs in the 2010 season -- and gathered an extensive amount of data on 16 teams.

To evaluate the teams as networks, researchers graphed player positions and ball movement among players, as well as shots taken. Then, they used that data to find out whether network metrics can measure team decisions in a useful way. The study involved more than 1,000 ball movements and 100 ball sequences.

"What that paper basically says is the 2010 data shows that the most successful teams are the ones that use a less predictable, more distributed offense and that connect their players more," said Fewell. "Those were the teams that had actually hired more elite players and allowed them to work together."

Fewell believes measuring team cohesion and communication is important.

"It's one way to capture the essence of a team and teams are everywhere for people," Fewell explained. "You work in teams for all different kinds of reasons. And the same, fundamental idea of cohesion and communication structure among the individuals and the team is critical and it's not easily quantified -- and this gives you a way to quantify it."

You may be wondering just how the Phoenix Suns measured up.

"I started working on this in part because I'm a Suns fan, especially of the 'run and gun' Suns," shared Fewell. "Our data suggested though that the 2010 Suns played the game as a fairly traditional point-guard centered play style. The Lakers and Celtics, in contrast, showed the network equivalent of the triangle offense, and it paid off for them. They were the teams in the finals that year."

The study included researchers from the School of Life Sciences and the School of Mathematical and Statistical Sciences, academic units in ASU's College of Liberal Arts and Sciences, and ASU's Center for Social Dynamics and Complexity.

Story Source:

The above post is reprinted from materials provided by Arizona State University College of Liberal Arts and Sciences. Note: Materials may be edited for content and length.

Journal Reference:

1. Jennifer H. Fewell, Dieter Armbruster, John Ingraham, Alexander Petersen, James S. Waters. Basketball Teams as Strategic Networks. PLoS ONE, 2012; 7 (11): e47445 DOI: 10.1371/journal.pone.0047445

Arizona State University College of Liberal Arts and Sciences. "Basketball as a statistical network." ScienceDaily. ScienceDaily, 11 December 2012. <www.sciencedaily.com/releases/2012/12/121211112719.htm>.
Arizona State University College of Liberal Arts and Sciences. (2012, December 11). Basketball as a statistical network. ScienceDaily. Retrieved December 1, 2015 from www.sciencedaily.com/releases/2012/12/121211112719.htm
Arizona State University College of Liberal Arts and Sciences. "Basketball as a statistical network." ScienceDaily. www.sciencedaily.com/releases/2012/12/121211112719.htm (accessed December 1, 2015).