There was a recent post on the NCAA website regarding the Moneyball philosophy - the idea that one can determine the value of sports players through the use of advanced statistical metrics - and what St. Bonaventure is doing to compete as a women's basketball program. (Ignore the misspelled title of the article.)
On the popular women's basketball message board known as Rebkell, this article caused a little bit of a kerfluffle. People tend to be all or nothing regarding the role of stats in basketball. Either:
a) They put far too much weight on numbers. They believe that PER and Usage and VORP are the be-all and end-all of everything. Before Moneyball-type stats came along, these were the same people that believed points per game and rebounds per game were the be-all and end-all. "Diana Taurasi must be better than Lauren Jackson, because she scores more points per game! Deny that, jerkface!"
(Disclaimer: I tend to fall too much in the "put too much weight on numbers" crowd. However, for someone who hasn't been following basketball for too long - just four years now - it's a good crowd to be in because the metrics given one at least the notion of which players are good and which players are not beyond the obvious.)
b) They deny that statistics can throw any light on the problem of evaluating basketball players. All of the limitations of these metrics are trotted out, and the hard-core anti-Moneyball crowd will take a trip into pure mysticism, singing the praises of moxie, pluck, grit, heart, clutch shooting and other chimeras of basketball.
As I've written before, stats are unfortunately used by some the same way a drunkard uses a lamppost - for support, and not for illumination. Stats are best used when casting light on dark areas, when the numbers reveal not answers, but questions. If a metric claims that unheralded Player A might be as good as - if not better than - superstar Player B, then one has two options: find the flaw in the data or look very hard at Player A, because it's possible that there might be something overlooked regarding Player A's talent.
So what is coach Jim Crowley doing? It's not as if he's trying to come up with the Grand Unified Theory of Basketball, some number he can use like the thumbs-up/thumbs-down of a Roman emperor at the Coliseum. Rather, the impression is that he's focusing on two things:
a) What are the basic aspects of a game most responsible for wins or losses? The article mentions that Jim Crowley wanted to bring down turnovers. Which is a good idea - turnovers are one of Dean Oliver's famous Four Factors. A team that can't hold on to the ball is going to have a hard time winning games. There are about a thousand other things a team needs to do to win, but not turning over the ball when you get it is very important.
b) There's a great quote from Crowley which illustrates the essence of Moneyball-like thinking?
"What if someone was small or not fast, but really had a basketball skill that we could utilize?" Crowley said. "If we found someone who could shoot, that was a basketball skill you can't replace - just like having a good eye in baseball. It would be on us to get someone into a situation where they could show what they were good at."
There was a great post on the Get Buckets blog about the imaginary "Five Tool" players of basketball. Namely, that there are five great skills a basketball player can have:
shot creation - essentially the ability to find a good look at the basket and shoot
The problem that schools like St. Bonaventure have is that they're not going to find players that are proficient in all of those areas. Those rare five-tool players are on their way to Connecticut or Tennessee. They get to fight for the three-tool and four-tool players.
However, without knowing much about advanced metrics, you can ask the right questions. Let's suppose you have Player A. Player A is good at creating a shot but she's a ball hog - she won't pass the ball - and she shies from contact, making it hard for her to pick up rebounds against more aggressive players. Player B is a great passer and rebounder, but can't shoot to save her life. Player C is a great lock-down defender but shaky in every other part of her game.
This leads to some questions:
a) When you're recruiting, which kind of player do you want? A, B, or C? Does the answer to this question change depending on the the position you recruit for?
b) If you can't recruit a five-star player, can you recruit one in aggregate? Maybe A, B, and C added together make one five-star player. If you know how to use player A, B and C - in what situations to bring them out, either individually or as a group - maybe you can be just as effective as a team with more talented players.
Is there a claim that Crowley has answered this question completely? I never read that anywhere in the article, and I suspect he'd deny such a claim. I'm surprised that free throw shooting was not mentioned, I'm aghast at how poorly some teams shoot at the stripe.
(Digression: I am a convert to the Oliverian religion of free throw shooting. In the little basketball computer game I play in my idle hours, I refuse to recruit a player who can't hit 70 percent from the stripe. I'll even disown computer players that do that; you can imagine what I think about real ones.)
Crowley might not have the answers to his questions - but I feel that he's asking the right ones. For example, Crowley is looking at total time-of-possession as a metric, feeling that the longer a team can hang onto the ball the more likely it is that they'll be successful. I don't know if this quite works for basketball, since some teams play at a quicker pace than others. But it would be a worthwhile project to see if time-of-possession correlates to winning games.
Undoubtedly, Crowley's outside-the-box thinking will offend some basketball purists. But I suspect that it will have a lot of fans. I'll be keeping my eyes on the Bonnies this year and see if Crowley has figured out the Grand Unified Theory that every coach is looking for. No one knows what the best method for finding the ultimate answer is; reading the numbers is just as valid as any other method.