A while back, I started to apply my passion for statistical analysis to our basketball team, referincing stats from kenpom.com, which I am hoping I am not stealing.
The primary trend in statistical analysis is to look at team stats in a tempo-free fashion. (This data is from ). For example, Buffalo runs an up-tempo attack. Accordingly, they led the MAC in scoring. But, does that mean their offense was the best? Conversely, a team that slows things down (such as Miami and BG), among others, won't score as many points as Buffalo, but does need to maximize each possession.
On the offensive, side, I think its fair to say that the contribution of these new metrics, was, well, meh. It didn't really tell us much that we did not know.
Let's see if defense fares any better. Here's what we concluded looking at the conventional stats.
- 9th in point allowed (note a couple big games inflate this number a little)
- 8th in 3FG% defense
- 10th in rebound margin
- 10th in steals.
- 12th in turnovers forced.
What does this mean. We were building this team on defense, and we did limit the opposition from shooting on us. However, they seemed to do a little better on 3s than on 2s, and that tells me that the inside D was good but the outside D (guards) was less good. Plus, teams got the benefit of more possessions, since we forced so few turnovers. Finally, once the ball hit the boards, they got some second chances, too.
- 5th in FG% defense
- 1st in blocked shots
The possession-based stats....well, they pretty much tell this story.
- BG was second in effective FG% (which factors in 3-pointers into the overall FG%).
- We were dead last in the percent of possessions in which we forced a turnover.
- We were 10th in allowing offensive rebounds.
- Were were in the middle of the pack in allowing free throws vs allowing FG attempts.
Its really the same story. Teams did not shoot well against us, but they got more attempts because of the low number of turnovers, and got lots of second chances. Really, not much knowledge gained here.
I think that if we are really going to break down this game, we can't do it just looking at boxscores and beating that data to death, multiplying here, creating a ratio here. You get the idea.
There are some really good questions to be answered, and I think possession-level stats would be really valuable. But data needs to be collected on each possession in raw form, not aggregated.
The same thing has happened in baseball--lots of new data sources have yielded new info on the game.
Some proposed questions:
- Does getting to the free throw line really help a team that shoots well?
- What players are on the floor when the team tends to score and defend well?
- How often does a possession with a blocked shot end up with an offensive basket anyway.
- Does the kind of turnover effect the damage? Does a moving turnover (like a steal) hurt more than a deadball turnover (like traveling).
- Are more offensive rebounds garnered on 2 point jumpers than 3 point jumpers.
- Are possessions with more passes really more effective?