What's the Suns' game plan? Ask the ASU scientists

Jennifer Fewell, a professor of life sciences, is a huge Phoenix Suns fan.
 

She watches every game, but lately it’s been in a distracted sort of way. Fewell can’t help studying how the players interact, how they move the ball across the court, who passes the ball to whom.
 

What else would one expect from a scholar who watches groups of ants and hives of honey bees for clues to their social networks?
 

The Suns have been a small team, and they didn’t have a dominant center, Fewell said, but they compensated by moving the ball quickly across multiple players. “I started thinking about how important team dynamics were to them and how that compared with what I’ve seen in the social networks I study.”
 

Her thoughts carried over to a presentation given in spring 2009 by Dieter Armbruster, professor of mathematics, who studies networks and complex systems (and who enjoys watching basketball but doesn’t root for any particular team), on “motif analysis” – examining networks via the connections among any three agents in the network.

“I realized this could be a way to tease apart group versus individual success in basketball teams and discussed the idea with Dieter,” Fewell said. “We were both excited about the prospect and started brainstorming how to go about it."

Dieter then advertised it as a possible project for students involved in an NSF-funded project called CSUMS (Computational Science Training for Undergraduates in the Mathematical Sciences) and two great students – John Ingraham and Alex Petersen – joined the group.
 

“I think it took both perspectives  – the behavioral and the mathematical – for us to see networks as a useful way to capture team dynamics.”
 

Fewell and Armbruster, both founding members of ASU’s Center for Social Dynamics and Complexity, began working on methods to analyze games, and eventually focused on the 2009-2010 NBA playoff games for their main database.
 

ASU’s newly minted “netball team” of Fewell, Armbruster, Ingraham and Petersen watched every play in every playoff game that season, keeping track of the ball movement from player to player. Each player was designated a “node,” and each ball movement was noted as a “link.”
 

Then, they evaluated ball possessions and found that there were 70 to 100 ball possessions in each game, each associated with an action. These possessions were all combined to generate a team “network” for each game.

From there, they began to identify the network characteristics for each playoff team, and how team play styles differed.
 
The next questions were, Armbruster said, “Can we say something about these networks, and is there a network that is characteristic of NBA basketball versus college basketball?”

Fewell adds a more practical question: “Can we use network dynamics to help understand what makes a successful team?”

The mathematical analysis of the playoff games yields a clear picture of the teams’ differing strategies – and a look at how their networks help them win or lose.
 

At opposite ends are the Lakers and the Jazz. 
 

“Phil Jackson is known for his triangle offense, and the team’s ball movements are much less predictable,” Armbruster said. “The Jazz’s network is star like, in contrast, with a much more predictable offense.
 

“The Jazz have low entropy – a mathematic feature that measures uncertainty – while the Lakers have high entropy. In general the teams with high entropy win against teams with low entropy.”
 

The study of team networks also has revealed that a shooter’s percentage may not be as important as it seems, Armbruster said. “We think that it’s not just how good a shooter is, but how good the sequence of plays is.”

Fewell adds, “The Lakers are particularly interesting because they manage to maintain high entropy  while still moving the ball towards the team members that are most likely to have a successful shot.”
 
Fewell said the study has suggested that how players operate as a team should be weighted against having marquee players, and that the teams that maintain a fairly consistent roster tend to do the best.
  

“This team cohesion, the ability to read other players and coordinate with them, is a large part of what makes basketball such a beautiful game. You can’t learn basketball just by practicing individual moves, because player contributions extend far beyond that.

“Grant Hill is an interesting example. Beyond his ball handling skills, he is also a key node in the Sun’s network. Successful paths to the basket often go through him; he is critical to distributing the ball.”
 

So what does it matter whether the point guard always passes to the shooting forward in one team, or another team’s center almost never passes it to the small forward?
 

And isn’t it a little frivolous for a mathematician and biologist to spend time analyzing the way a sport is played?
 

Since April is national Math Awareness Month, it’s a good time to point out that Armbruster and Fewell’s study indeed has value far beyond the fun of making mathematical “maps” of your favorite NBA team.
 

Armbruster, who has long been interested in network dynamics, said learning how networks function is extremely valuable “when you are trying to control or engineer complex systems, such as electric power grids.
 

“For example, if a transformer explodes, the power failure can cascade through the whole network. We need to understand what is in the network structure that makes it so susceptible to these failures.”
 

Studying travel networks – how often and when people drive or take public transportation to various destinations – also can help authorities plan for outbreaks of flu or other illnesses, Armbruster added.

 
“If you know how a virus would likely spread through the country, you can how much benefit there would be from shutting down an airport, for example.”
 
Studying networks also can help researchers understand better how foreign species such as the zebra mussel are likely to spread through national waterways, Armbruster said.
 

Beyond applying the study of basketball networks to larger social issues, Armbruster joked that a more careful analysis could “improve the odds makers in Las Vegas. We could determine whether certain styles of play are more successful than others, such as the star vs. the triangle.”
 

After her marathon of watching playoff games as a scientist, not a fan, Fewell said she had to take a break from her favorite sport. “After that I couldn’t watch basketball for a few months.”  However, she notes that she has now recovered, and that she and Armbruster are busy analyzing networks for the recent NCAA Sweet Sixteen.