Friday, June 1, 2012

No Offense Intended

There's an old adage in basketball that goes, "Defense wins championships." I have no idea if it's true. My hunch is that this is just a clever way of motivating young men (and women) who are much more interested in putting the ball through the hoop than squaring up their defender to put some effort into playing D. In truth, I suspect that good defense coupled with good offense wins championships. If you take a look at Ken Pomeroy's offensive and defensive rankings of college teams over the last decade, you will find that it really takes both.

I want to explore this idea a little bit here, although before I get going I should note that the data I have at hand at the moment are hardly ideal for the task. I want to examine the average number of points scored and allowed per game by Arizona basketball teams by season from 1979 to 2012. Then, I will compare each of those numbers to win percentage to see which factor contributed more to the final record for the season.

The graph above shows the average points scored per game by Arizona teams (in black) and their opponents (in red) over the last 33 seasons. I want to first point out a few things. From the 79-80 (coached by Snowden) through the 83-84 season (coached by Olson), our opponents scored more points than we did. This situation was rectified in Lute Olson's second season as coach and he never looked back. Since 1984, there has only been one season in which the Cats gave away more points than they scored, and that was only three seasons ago(2009-10), Sean Miller's first year as head coach. Still, the Cats finished that season with a winning record of 16-15.

It is also worth noting that to some degree points for and against track each other. Teams that score a lot of points tend to allow more points to be scored. This is because there is an underlying factor that governs both variables, and that is tempo. Teams that play quickly have more possessions in which to score points and to allow points to be scored. Teams that play slowly will be characterized by higher point totals on offense and defense. At the moment, I can't control for tempo over this time scale, so we'll largely ignore it.

Over this time period the Cats have averaged between 62.4 and 90.8 points per game on the offensive end, the lowest average coming again in the 83-84 season and the highest in the 97-98 season, the year following the National Championship. The range on the defensive end spans 59 to 78.7 points. Amazingly, these two point totals occur only two years apart. The high number comes from the transitional season between Snowden and Olson when Ben Lindsey coached the team (1982-83). The low number is from Olson's 2nd year as coach in the 1984-85 season. As can be seen in the graph above, while it took Olson a few more years to get the offense humming, he turned the defense around very quickly, at least using this coarse measure. Similarly, Sean Miller has lowered the number of points allowed by his teams by nearly 10 points over his first three seasons.
The graphs above show the relationships between winning percentage (y-axis) and point differential, points for, and points against (x-axis). Not surprisingly, the strongest relationship present is between point differential and winning percentage. What may be more of a surprise is that points scored is a much stronger predictor of winning percentage than points allowed. In fact, the relationship between points allowed on defense and winning percentage is not statistically significant, meaning that the relationship is no better than expected for completely random data.

At face value, these data would suggest that offense wins championships rather than defense, or at least that offense wins games. That statement, however, should be taken with a huge grain of salt. Points allowed per game is not a very good measure of the defensive efficiency of a team. Last season, the top six teams in this category were a mix of very good and so-so defensive teams (Wisconsin, Virginia, Stephen F. Austin, Harvard, Drexel, and Wyoming). Of course, the same could be said of points scored (except that Kentucky is No. 11). Nonetheless, at least for the Cats, scoring more points has lead to winning more so games much more than allowing fewer on the defensive end.