Over at Grantland, Zach Lowe wrote twoexcellent pieces about the way the Toronto Raptors’ stat analysts use SportVU data. Long story short: SportVU is a system of six cameras that films every action that takes place on a basketball court. The Raptors’ analysts take the SportVU coordinate data and generate a diagram of the ball and all ten players on the court. With input from Raptors’ assistant coaches, they have also created an algorithm that demonstrates ideal defensive positioning, and can compare this ideal to the actual positioning used by the players on the court. The piece actually includes three video examples of the diagramming at work, along with some more interesting tidbits — it’s definitely worth a read.
Of course, some basketball writers have used this article as an opportunity to question the use of statistical analysis in basketball. For example, here is Doug Smith of the Toronto Star:
As you know, I’m a tad skeptical – I believe advanced analytics have a place in a very human game, I’m just not sure how big that place should be – but the thing I worry about is it driving some kind of wedge between coaches, general managers and the analytics guys who truly believe their analyses are correct.
…There is enough “creative tension” in many GM-coach relationships that an added level could be the one that causes and irrevocable split.
…Any workplace – and I don’t care if it’s widgets or the NBA – works better when the leaders have the respect of those they are leading and if the personal relationships can make a group better than the sum of its parts. And no workplace – and I don’t care if it’s widgets or the NBA – works as well as it can if any key components are at cross purposes.
If this “added level of creative tension” is enough to cause an “irrevocable split” between a team’s GM and its coach, then perhaps the team is employing the wrong coach. All well-run businesses use statistics and numbers to inform their decision making process. Even food and beverage companies — not usually the type of companies we normally associate with cutting edge statistical analysis – use statistics to inform their decisions. When a beverage company wants to boost its sales, they employ the services of someone like Howard Moskowitz. Moskowitz, who has studied math and has a Ph.D in experimental psychology, uses the process of “product optimization” to determine the ideal composition of a food product. From the New York Times:
In the process of product optimization, food engineers alter a litany of variables with the sole intent of finding the most perfect version (or versions) of a product. Ordinary consumers are paid to spend hours sitting in rooms where they touch, feel, sip, smell, swirl and taste whatever product is in question. Their opinions are dumped into a computer, and the data are sifted and sorted through a statistical method called conjoint analysis, which determines what features will be most attractive to consumers.
CEOs and decision makers at food companies know that statistical analysis and the scientific method are the best ways to make informed decisions. This is because it works. Any company that doesn’t embrace these processes may experience short-term success, but in a competitive market they will always end up losing in the long term.
Sports organisations — including professional teams — are no different. The ones that commit to evidence-based decision making will succeed in the long term. If an organisation that is trying to be evidence-based hires a coach who doesn’t “believe” or “trust” statistics, then clearly this organisation is not fully committed to evidence-based decision making. If you want your organisation to embrace statistics, don’t fill key positions with people who don’t share your philosophy. If you both use statistical analysis and employ coaches who don’t embrace stats, your approach is not only incoherent, it’s also contradictory. And finally, when you both use statistical anaylsis and employ coaches who are vocally opposed to the use of statistics, outsiders know that your commitment to evidence-based decision making is superficial.
If, by trying to implement evidence-based decision making, an organisation creates an irrevocable split between it and its coach…too bad for the coach. The coach and the organisation are clearly not in agreement on an important issue, and so the coach should be let go. The sooner this disagreement is recognized by the organisation, the better. In this respect, Doug Smith is correct.
There’s another issue related to Zach Lowe’s article as well. Some people believe that, in allowing Zach Lowe access, the Raptors’ stat guys have tipped their hand and revealed amazing secrets. Cathal Kelly, who also writes for the Toronto Star, is one such person:
Fourteen teams decided that sort of information is best kept to themselves. One team decided to pull back the curtain and let ESPN see Oz at work.
…Three animations accompany the text. When told that on Tuesday afternoon, Raptors coach Dwane Casey reared back in surprise.
“What?” Casey said. “Not the actual thing?”
The actual thing.
Casey was left briefly speechless. It is difficult to credit that the people guarding the team vaults are also doing guided tours.
Of course this idea is preposterous. What else would you do with a bunch of coordinate data, write a song? What the Raptors stats guys did with the coordinate data — as awesome as it is — is blatantly obvious and the first thing that comes to everyone’s mind when they hear about SportVU. The trick is to actually build a program that will turn the data into something usable. Likewise, demonstrating that they’ve created an algorithm to determine ideal defensive position is not going to give very much away, as they have not revealed how the algorithm actually works. To continue the food example, it would be like people saying Howard Moskowitz has ruined his livelihood by revealing that he uses conjoint analysis when he optimizes food products. Which is stupid.
As they say, the devil is in the details.