Andy Katz recently published his preseason top 50 for the upcoming season. He did a very thorough job and I think captured something close to what the preseason polls will have. Although I am a little puzzled as to why Gonzaga is ranked as low as 22. At any rate, it’s time to start speculating on the upcoming season.
Part of doing that is accurately assessing how good a team was last year. Some opinions will be inflated and some deflated based on the distribution of luck. Can we quantify which teams were lucky and which teams were unlucky last year? We can take a shot at it. (Why do I keep saying we? I am sounding like Beano Cook here.)
Bill James can help us. He’s a genius who unfortunately has applied his genius to Major League Baseball. He was the one who invented the Pythagorean Theorem of Baseball, which looks like so:
Expected Winning Percentage = RunsScored^2 / (RunsScored^2+RunsAllowed^2)
This is similar to the more famous Pythagorean Theorem which is used to determine the length of the hypotenuse of a right triangle. What one has to do with the other is anyone’s guess. But there was also a time when chocolate and peanut butter were not thought to be a good match either.
I’m not going to get into why it works, because I am not exactly sure myself, but it works very well. The basic idea is that the more you outscore your opponents over the course of the season, the more games you should win.
This is one way to determine how lucky a team is. To buy into this, you have to believe that the outcome of close games is more due to luck than the outcome of blowouts is. It seems to be an obvious point, but there are those who still believe the Earth is flat also.
If a team wins a lot of close games and loses a few blowouts, their actual record would exceed their “Pythagorean record.” They would be the beneficiary of luck, because you would not expect them to be dominant in close games especially given that they had a few lopsided losses. The opposite case would indicate a team is unlucky.
The formula’s ability to predict a major league team’s record is uncanny. It probably works well because baseball teams play a similar schedule. This isn’t true in college hoops, but that problem can be minimized if we limit the formula to conference games. Then we can determine how lucky a team was relative to its conference.
The formula has to be modified to apply to basketball – instead of squaring points scored and points allowed, they need to be raised to the 10th power. In doing that, the formula also has an uncanny ability to predict the conference record of a given college hoops team.
Though for a few teams it does poorly. These are the teams that presumably were affected by an unusual amount of luck. Later this week, I will post the teams that benefited or were hurt the most by luck last season.