With the combination of the expectations surrounding the Oregon basketball program, and the absence of an NBA season, I expect more ATQers to be watching college basketball than ever before. Since the release of Moneyball, advanced statistics in baseball have gotten a lot of attention, and that has followed in football in recent years. However, advanced statistics are alive and well in basketball as well, and my aim is to provide a brief but informative introduction to those who have not used these before.
The godfather of advanced statistics is Ken Pomeroy, whose site,kenpom.com, is the holy grail of basketball databases (unfortunately, he has decided to charge $19.95 for use of his database this year, but its well worth it if you are a basketball junkie). He gets into some fairly advanced stuff, and I don't want to get into them here. For example, Pomeroy rates every team, but the system is a bit complex, and I'm not sure that I understand all the ins and outs. Using this, he also predicts all the games (He predicts Oregon at 18-12 on the season, 9-9 in the PAC). Perhaps we'll get into this later in the season, but I want to start with more basic figures.
The most frustrating thing about traditional basketball metrics is that they lack context. Averaging 20 points per game is different in a run and gun offense than it is in a Princeton offense where you want to slow it down. Field goal percentage means a different thing if you're a big guy who scores mostly on dunks than if you are a guard shooting a lot from three (you may make it less, but you get an extra point when you do). We must attempt to find stats where we can directly compare teams without the noise. Pomeroy's most relevant statistics are what he calls his "four factors," which give you a good indication of the two things that are most important in basketball. How good are you at generating more possessions than the other team? And how good are you at converting possessions into points? The teams that are better at these four factors usually win the game.
Effective Field Goal % -- As noted above, the problem, the problem with traditional field goal percentage is that you don't need to make as many threes as you would twos to score the same number of points. eFG% takes that into account by putting a 50% premium on made three pointers:
eFG% = (.5*3FGM + FGM) / FGA
Turnover percentage -- Measuring turnovers is a great way to measure which teams get extra possessions, but the problem is with pace. Its almost guaranteed that Kentucky will have more turnovers per game than Oregon State because Kentucky runs a wide open system that puts a premium on speed and will mean a faster game. Turnover percentage is designed to be pace independent:
TO% = TO / Possessions
Offensive rebounding percentage -- Again, the traditional metrics of rebounds per game don't take into account pace or how good of a shooting team you are in the first place. Offensive rebounding is another way to generate extra possessions:
OR% = OR / (OR + DRopp)
Free throw rate -- We often use free throw percentage, which is important, but a team that hits 70% but gets to the line 30 times is more effective than a team that hits 90% but gets to the line 10 times. This is a measure of how effective teams are at getting to the line:
FTRate = FTA / FGA
While the abolute value of thee four factors aren't completely equal, put together they place the highest predictive value of success of any set of basic statistics, and are far more relevant than any traditional metrics. You will see them as a cornerstone of ATQ's basketball coverage.