Interesting point, Ray. If I understand right, you’re talking about the situation now where a point guard is expected to carry more of the scoring load than he used to. I’m sure you’re right that the differences in the job description–what was expected of a point guard–would require different methods of evaluating them over time.

Given this I can really see the appeal of RT’s suggestion of using each individual player’s +/- score conditional on the +/- scores of the other players on the floor. Certainly this approach has its flaws, but it’s good in that it looks at the outcome of interest–team points–without getting bogged down in potential distractions like assists.

I think that’s why some of the most famous point guards over the last two decades are from teams that ran specific offenses that depended heavily on coordination, timing and floor management. Magic Johnson and John Stockton come to mind immediately, and it’s instructive that the new “floor generals” of note play for . . . the Utah Jazz (Jerry Sloan) and the Charlotte Hornets (Byron Scott). That’s no coincidence.

Just a thought.

Interesting discussion about basketball stats–so where should I start if I want to read what the basketball statisticians have to say, RT? I’m at least fairly well read about baseball, but I don’t know the first thing about statistical analysis in basketball.

Mark, I agree with you about liking baseball because the stats are the most open. The discrete nature of the play really contributes to measuring performance meaningfully. You can do all kinds of fun things like win probability analysis that I think the Hardball Times people do, where you estimate the probability of each team winning the game before and after each play, assign credit/blame for the play, and you have a measure of performance that is completely situation-adjusted, at least within the game. Of course it doesn’t account for the meaningfulness of games–it assigns the same value to a game if winning the division hinges on it or if it’s the last game of the season for a team that’s going to finish with over 100 losses.

To the degree that analysis of sports performance can be broken down into measurement versus statistics, I lean toward enjoying the statistics side of things. I’d rather have measurement questions settled and have good measures that I can use to answer fun questions, although I can definitely see the appeal of asking interesting questions about measurement. I think that’s probably why baseball is my favorite.

Stats are flawed in a game like basketball because there is such an interplay between what one player does and how that influences another’s stats. My favorite example is Jordan and Pippen. Pippin was seen as a Hall of Fame player and one of the best 50 of all-time when he played with Jordan. Take Jordan away, and Pippin was just another all-star.

The real test of ability is how someone performs when they are the only option – when their team is a waste of space around them. The Cavs getting to the finals last year was amazing. You can beat the Cavaliers by triple teaming James; you can’t beat the Celtics by triple teaming Garnett – or the Jazz by triple teaming Boozer. I can’t think of a single NBA championship team that didn’t have at least two legitimate all-stars, if not two legitimate Hall of Fame players. (There probably is one, but I don’t care enough at the moment to go back through all of them. Even the Lakers this year only are what they are because they have Fisher and Gasol.)

Batting average is one measure of success, but slugging percentage is even better because it accounts for multiple base hits. And OPS is even better because it accounts for multiple base hits, plus bases on balls, hit by pitch, sacrifice flies, etc. (One reason I love the game is because sometimes making an out (via sac. fly) is nontheless considered good for the team because it advances a runner, so consequently it counts in the offensive stats.

The downside is that you can get carried away with them. A few years ago I got really excited about batting average with runners in scoring position. Some batters were just average .275 hitters until they had teammates on second or third, then their average jumped to .450. They then started to measure BA with the go ahead run on second or third in the late innings. I fell for it hook, line, and sinker and became an annoying shill until it was demonstrated to me that BA is remarkably stable over time, and that any observed differences with runners on base are anomalies that even themselves out. Then I felt pretty dumb.

Hmm. I’ll take your word for the skepticism about offensive stats. I will maintain, however, that team shooting % is still a more valuable tool for evaluation than numbers of assists.

I agree completely with you about team defense. Larry Bird had slow feet and was therefore a horrible one on one defender, but he was invaluable on team defense, because he understood when to pressure the ball, when to double team, and when to sag back into the passing lanes. Some players have an almost uncanny knack for coming up with a steal at a critical moment in the game.

In general, I think stats are fun if they help explain what happened in the game. Baseball is the most open, because you can look at a box score and make sense of what happened, even if you didn’t see the game. Football is a mess because the stats don’t reveal much, and sometimes they actually obscure what happened. Basketball and hockey are the most fluid of games, so it is hard to capture what is going on with numbers.