Bull Durham is one of my all-time favorite movies. It was Kevin Costner before every role he played was just him being Kevin Costner (the same thing happened to Harrison Ford).  One of my favorite parts of the movie is when Skip blows a gasket in the locker room:

This is a very simple game. You throw the ball. You catch the ball. You hit the ball. Sometimes you win. Sometimes you lose. And sometimes it rains. Think about that for a while.

I love it because it's true. Oh, I am not saying that baseball is easy. In my opinion one of the first lessons life should throw anybody is that simple and easy are not even close to the same thing.

Lately, I've been talking to people a lot about how we think about the game of basketball conceptually, and I have come to the conclusion that basketball is also simple (but not easy). Dave Berri gave it a good start the other day:

Teams win because teams are able to

• gain possession of the ball (i.e. grab defensive rebounds and force turnovers)
• keep possession of the ball (i.e. avoid turnovers and grab offensive rebounds)
• ultimately turn possessions into points (i.e. shoot efficiently)

I think I would add a fourth bullet point: "keep the opponent from turning possessions into points (i.e. make them shoot inefficiently)". Another way of putting this is to say that Basketball is a Simple Game:

• You get the ball.
• You keep the ball.
• You make your shots (note how this is not "You score a lot of points". c.f. the 1991 Denver Nuggets).
• You make sure the other guy doesn't do any of that (well).
• ????
• Profit

And of course, just like baseball, it's simple but it's not easy. It starts with the fact that there are multiple ways to do each of these. You can get the ball by grabbing defensive rebounds, or picking off passes, or slapping the ball out of a players hands. Or taking a charge, or....you can see how a 3 or 4 column matrix suddenly expands to a lot of columns.

But really, this is what statistical models are about; reducing things to common elements, to variables you can control for. And although there's a lot of math involved, it really all boils down to how each box score affects winning. If it doesn't correlate with wins, then the model cannot use it. Which at times leads to shortcomings in the model. And this is why Wins Produced explains 95% of team wins; the other 5% is "hard" stuff like team defense that we haven't figure out how to model yet. But then again, I find this preferable to other models that just lump a bunch of numbers in, even though those numbers have no correlation with winning, and then one closes their eyes, does some handwaving about "intangibles" and "team chemistry", and hopes that somehow the model makes sausage instead of spam.

Get the ball. Keep the ball when you have it. Make your shots.

Simple!