Filed under:

# What's in Store for Dumpy's Statistical Analyses during the 2006-07 Season

Hello everyone, and welcome to another season of Dumpy’s Statistical Analysis. For the record, I’m Dumpy (not my real name), and what I try to do in this space is to take a numerical approach to analyzing the performance of the Nets and the individual players. I’m not a statistician by trade, but I love crunching numbers and detecting behavioral patterns, and so here we are.

There has been an explosion of basketball-related numerical data in recent years, such as John Hollinger’s work and that of the web site www.82games.com. However, we won’t blindly follow their analysis. Some of these other sources "hide" their formulas, and just present their data without context. What does it really mean that Jason Kidd has a "Hands" rating of 31.8? Is this good? How is this number derived? I have no idea, and, as far as I can tell, 82games.com doesn’t tell us anywhere on their site. In these columns, I try not to simply present numbers, but to explain where they come from and what they might mean. There’s no "magic formula" that can tell us, in one number, what player is better than another or why the team is successful. Numbers are just tools that inform our analysis, but they can’t replace it. That’s why it is imperative that we understand exactly what the numbers represent.

Last season, I spent a lot of time analyzing the plus-minus figures earned by each player and combinations of players. The plus-minus rating is a way to measure each player’s combined offensive and defensive total contribution to the team. Two simple examples will show how this works, taken from a game against Miami last February 4. In that game, Vince Carter was on the floor for 40.0 minutes, during which the Nets outscored Miami by a total of 21 points—so Carter earned a +21 rating. In Jacque Vaughn’s case in that same game, the Nets were outscored by the Heat by 10 points during the time he was on the court, so he "earned" a -10 rating. That’s all there is to it. Sometimes, you’ll see the plus-minus figure incorporate data of how the team did when a player was NOT on the court, but I don’t think that is too significant, since it essentially just compares the starter’s performance to that of his backup.

We can also measure plus-minus based on two-man combos, three-man combos, 5-man combos, etc., which can enable us to identify groups of players that play exceptionally well or poorly together. Data on 5-man combinations can help us identify the players that contribute the least or the most to team success by comparing different combinations that differ by one player. For instance, if we know that, over the course of the season, a unit consisting of [Kidd, Carter, RJ, Collins and Krstic] has outscored its opponent by a greater margin than a unit consisting of [Kidd, Carter, RJ, Robinson and Krstic] in the same number of minutes, then we can generally conclude that Collins adds more to the team success than does Robinson. However, we have to be careful not to make a sweeping generalization that Collins is better than Robinson; there may very well be other four-man combinations that Robinson plays better with than Collins does—it is just up to us to identify them. A web site called www.popcornmachine.net provides plus-minus figures on a game-by-game basis, and I rely on these figures.

I also spent some time measuring the team’s proficiency at rebounding by measuring the number of offensive rebounds relative to the number of opportunities. It turned out that the Nets were one of the best teams in the league at restricting the opponent’s offensive rebounds, but one of the worst at getting offensive rebounds themselves. From time to time, I also looked at how well the Nets and their opponents did at converting those offensive rebounds into points.

This season, I’ll continue to attack those questions, although with less detailed textual analysis and description in an effort to complete analyses on a more regular basis. In addition, I’m planning on adding some new features, which are based on the work of Dean Oliver, as set forth in his recent book, "Basketball On Paper."

Oliver describes methods for determining the efficiency of a team’s offense, as well as that of individual players. This season, we’re going use these methods to measure the efficiency of the Nets on a game-by-game basis. Here are some of the statistics that you’ll see:

Possessions

. Basically, possessions are the number of times a team brings the ball up court. Counting the number of possessions in a game is a good way to measure the pace of the action. For games involving running or trapping teams, the number of possessions will be high—possibly more than 100. For more methodical teams, the number of possessions may be closer to 80. Regardless, the number of possessions will be approximately the same as the opposition in each particular game. Possessions can (generally) end one of three ways: on a field goal attempt that is not rebounded by the offense (this includes successful FG attempts); on a turnover, or through some free throws. Even though the NBA doesn’t count actual possessions, it can be estimated using those stats as they appear in the box score. You’ll see that we’ll also present the "average" number of possessions for both teams for better accuracy.

Offensive Rating

. A team’s offensive rating is just the number of points scored per 100 possessions.

Defensive Rating

. A team’s defensive rating is just the number of points scored by the opponent per 100 possessions.

Assist Percentage

. This one is easy. The Assist Percentage measures the percentage of successful field goals that have been assisted.

"Big Four" Factors

. Through his research, Oliver has discovered that there are four primary factors that determine the outcome of a basketball game: field goal percentage, offensive rebound percentage, turnovers, and the ability to get to the line and hit free throws. We’ll refer to these as the "Big Four" factors, and will measure them for both the Nets and their opponents. Offensive rebound percentage is measured as indicated above, as a percentage of opportunities. Likewise, turnovers will be measured as a percentage of possessions. We’ll measure free throws by measuring the percentage of time the team got to the line in relation to field goal shot attempts.

Scoring Possessions

. This is an estimate of the number of times a team scores at least one point on a possession.

Field Percentage

. This is an estimate of the percentage of times a team scores a basket on possessions where no free throws are awarded. This illustrates how well a team scores when not drawing fouls.

Number of plays

. This is an estimate of the number of times that a team both gains control of the ball and when they lose control of the ball, either when the opposing team gains control or when a shot goes up. The number of plays will be greater than the number of possessions. For example, if a team misses a shot, gets the rebound, and then makes the follow-up, that would count as two plays and only one possession.

Play percentage

. An estimate of the percentage of a team’s plays on which it produces a scoring possession.

In addition, we’ll use many of these same concepts to look at the efficiency of individual players within a game. These are the source of the individual "Offensive rating" numbers that can be found places such as www.basketball-reference.com. Applying these concepts to individuals is a complicated process, and, if I may say so, Mr. Oliver could do a better job at explaining them, and never addresses the error rate of the "points produced" stat (which we’ll address next time). Nevertheless, we’ll try to use these concepts as best we can, and in the next Analysis, I’ll try to tackle exactly what they mean and what they measure.

Let me add, too, that Oliver has devised a method of calculating an individual’s "defensive rating," which you may have heard of. Problem is, the data required to compute this figure is not recorded by the NBA. Oliver does the best he can, using a series of assumptions, but I have concluded that the number and scope of these assumptions really limit the value of this statistic. Upshot: You won’t see it here.

So this will be the typical format of one of my Analyses this coming year, and from time to time I’ll tackle a more challenging question as the mood strikes. Also, from time-to-time, we’ll provide year-to-date figures for these measurements, and compare them to those of other teams so you can see how the Nets stack up.

So, I hope to see you around. Have an enjoyable season! And don’t forget to write.

--Dumpy.