Wednesday, November 14, 2007

What do these stats mean?

Just as a matter of clarity, I figured I would ensure that I had a good reference available for all the ratings I use for Duke over the course of the season.

Floor Percentage

Floor% is calculated as personal scoring possessions divided by total individual possessions. The idea is that it tells you how reliable a scorer a player is. Often big men have higher floor percentages, while scorers/outside shooters have higher efficiency (because of the high number of possessions they have with 3 points). Scoring possessions include actual field goals made and free throws, while individual possessions also include missed shots and turnovers.

Individual Efficiency (or offensive rating)

It's a complicated formula that tells you the points a player produces per 100 individual possessions. I say "produces" because it's not just a matter of scoring, but it also takes into account assists and offensive rebounds. For instance, in spite of the fact that DeMarcus Nelson was Duke's leading scorer in 06-07, McRoberts actually produced more points per game (13.88 to 13.13) because of his 20.63% assist rate as compared to Nelson's 13.93% assist rate.

Usage Rate

Usage rate is basically a calculation of what percentage of the team's possessions a player "uses" when he's on the floor. A player "uses" a possession when the possession ends with something he does - whether that's turning the ball over, making a basket, shooting free throws or missing a shot that isn't rebounded offensively. With 5 players on the court, a perfectly balanced team would have each player with a 20% usage rate, and not-shockingly, the majority of players land within the 18-22% range. Players who use more than 25% are legitimate stars who carry a huge burden for the team (some examples: Durant used about 32% in 07, JWill used about 31% in 01, Redick used about 29% in 06, etc.). The combination of usage rate and efficiency over the course of 4 years or over the course of a season can tell a lot about a player. Some players can use a lot of possessions pretty efficiently (for instance, Durant used 32% of possessions in 07 with a 117 efficiency), and others can carry a smaller load very efficiently, but has trouble if he carries a bigger burden. This way, we can determine what the best "roles" for individual players are based on where they can maximize their offensive efficiency.

Stop Percentage

This is a calculation of stops per possession. A stop is when a defensive player ends a possession without the other team scoring. Calculating this stat requires a number of unconventional stats like forced misses (not blocks), forced turnovers (not steals), defensive field goals made, defensive free throws made, and forced free throw misses.

Individual Defensive Efficiency (or defensive rating)

Basically a calculation of points given up (from both free throws and baskets) divided by total defensive possessions (which includes misses, turnovers caused, etc). As I've noted, this rating has a number of limitations. For instance, a player who is a team's main defender will often get the hardest assignments - and might give up more points than a player who is guarding a lesser player. This is probably more a concern in the NBA than at the college level because there are fewer 25-30+% usage players in the college level. In fact, I imagine Duke will only run into Hansbrough as major guys who use over 25% of possessions regularly.

Defensive Usage Rate

The estimated defensive ratings (based on conventional stats), assume a 20% usage rate for each player. However, using more elaborate defensive stats, we can determine which players are most involved on the defensive end. Not shockingly, the two most involved players in Duke's first two games were Zoubek and Singler respectively. The nature of the game usually means that help defenders will often be involved in defensive plays while on the floor. One interesting note for usage rate is that it might suggest something on its own when compared with a player's replacement. In other words, in Duke's first two games, Paulus has had a defensive usage rate of about 22% and 31%, while Nolan Smith's in the same period has been 6.3% and 1.7%. To some extent, that is affected by the fact that Paulus has caused more turnovers, but also because players have shot fewer shots over Smith than over Paulus, in spite of the two of them playing the same position and guarding the same players. Could that suggest that Smith has been in superior position to prevent his man from wanting to try to score, whereas Paulus (for any number of reasons: less quickness or more gambling for turnovers) has defended many more shots because his man seems more aggressive? Maybe. If this becomes a trend for the season, it might tell us more.

Others

There are a number of other per-possession stats that I use pretty freely here, but they are easier and more intuitive. Defensive rebounding rate is simply the percentage of possible defensive rebounds a player or team grabs. For players, it's the same calculation but the total rebounds available (that is, the sum of team defensive rebounds and opposing team's offensive rebounds) is multiplied by the percentage of the minutes they have played. Offensive rebounding rate is the same idea. Assist rate is the percentage of baskets a team scores that an individual assists on while he's playing. Block rate is the percentage of the other team's 2-point baskets a player blocks. Steal rate is the percentage of opposing team's possessions that a team or player steals the ball on. Turnover rate is the percentage of a team or individual's possessions that ended in turnovers.

Finally, I have entirely replaced FG% with eFG%, which is a calculation where half of the 3 pointers made are added to the total field goals made. In other words, it's a calculation to try to normalize FG% to points scored rather than shots made. If I shoot 10 3 point shots and make 4 of them, I've scored 12 points on 40% shooting. If I shoot 10 15' jumpers and make 6 of them, I've scored 12 points on 60% shooting. It makes more sense to normalize the FG% so that we can tell how well individual players can turn shots into points. Therefore, 40% 3-point shooting is equivalent to 60% 2 point shooting.

No comments: