Tuesday, February 3, 2009

What Does It Take to Win in Basketball?

As we are right in the middle of an especially interesting NBA basketball season and the always interesting college basketball season, I have been intrigued by the various and sundry basketball announcers, virtually all of whom tell us that to win a championship, you have to play great defense.

In considering this subject, I took a look at Basketball on Paper by Dean Oliver, which is probably the foremost guide to statistical analysis of basketball and the author of that book notes that he believes the phrase "Defense wins championships" probably goes back to the titantic Boston-Philadelphia and Boston-L.A. NBA play-off series of the 1960's.

The media leaped on defense as the simple explanation for what was happening since Russell was great at defense, and Chamberlain, West and Baylor were great at offense.

In reality, things were a lot more complicated. But what does stand out from my study of the NBA standings is that adjusted point differential or SRS, seems to be the greatest predictor of success in the NBA play-offs. Because, as in college ball, NBA teams have not played equal schedules, the records are often misleading.

For example, the 1969 Lakers, who finished 55-27 and first in the West, are often chided for losing to the 48-34 Celtics, who only finished in fourth place in the East. But guess what? Due to statistical anomalies, Boston essentially tied for first in the NBA that year in SRS, finishing at 5.38, while L.A., which won seven games more, only had an SRS of 3.84. Boston ended up beating L.A. in the finals in a close seven game series.

Celtic fans try to have it both ways with the Celtic teams of the late 50's and 60's. They try to say that they were both the best teams of all time and that they were incredibly clutch. I would say that it is much more true to say that they were the best teams of that period by far and that they managed to avoid being upset by inferior teams, although often just by the skin of their teeth.

Out of the 13 years Bill Russell was with the Celtics, the Celtics had the best SRS in the league 11 times(including 1969 which was essentially a tie with the Knicks), and guess how many championships they won?

If there is a critique to be made of those Philadelphia teams with Wilt and those Laker teams with Baylor and West, and later with Chamberlain, it is not that they lost to Boston in the play-offs. That was the expected outcome. It would be why couldn't these teams match Boston in the regular season.

As regards Wilt Chamberlain, he played on four teams that led the league in adjusting scoring differential, in 1967, 1968, 1972 and 1973. Thus, Chamberlain played on four teams that would be deemed the play-off favorites and he ended up winning twice, in 1967 and 1972, although the 1972 Knicks were only a couple of points behind the Lakers during the regular season.

Adjusted point differential (SRS) is simply a team's offensive average minus its defensive average, with an adjustment upward or downward based upon schedule strength. Among power ratings in basketball, this is basically what Sagarin in USA Today calls "pure points" adjusted for schedule strength and I would bet that if we look, we will find that teams that win the NCAA title generally have excellent point differentials. Sagarin indicates that pure points is not politically correct but does the best job of predicting success.

Generally, in the NBA, an SRS of over 8 points is outstanding and will almost always result in that team winning the title. As far as I can discern, no team has ever led the league in SRS at an average of 9 points or more, without winning the title. This select group includes the 1971 Jabbar/Robertson Bucks, the 1972 Chamberlain, Goodrich, West Lakers, the 1986 Walton/Bird Celtics, and three different Jordan-led Bull teams.

Because expansion, injuries and improvement throughout a season can affect team success, SRS is not the final arbiter of greatness, but it certainly is highly correlated.

Only three teams have exceeded 11.5 points: the 1971 Bucks, the 1972 Lakers and the 1996 Jordan/Rodman Bulls, and all three won 66 games or more during the regular season and went through the play-offs at a clip of .800 or above.

The 1972 Lakers, who went 69-13 in the regular season had to play the almost equally worthy Bucks that year in the Western Finals. The Bucks went 63-19 and had an SRS of 10.70, which is one of the five highest of all time, but went down 4 games to 2 versus the Lakers.

So, it might be worth thinking twice before simply accepting the old "defense wins championships" platitudes. The defense versus offense paradigm may not, in fact, be especially illuminating. It may be useful for coaches in making team adjustments or for discovering why a particular game was won or lost, but it may not be particularly informative in terms of telling us which team is better, when an "offensive" oriented team faces a "defensive" oriented team. I doubt any of this will stop the television announcers from making the claim, however.