Bucks Diary

Tuesday, April 08, 2008

Win Splits for every NBA team's offense and defense


How many victories does each team's offense and defense contribute to their respective win totals? I investigated this question using a derivation of the Win Score metric developed by the authors of "The Wages of Wins".


The Logic being used, and How to interpret a "Win Split"

My theory is that 50% of each team's win total can be traced to the team's overall Win Score per game average as compared to the overall NBA Win Score per game average (which is 43.3 Win Score points per game). This I consider, in a slightly dirty way, the team's "offensive" output. The other 50% can be traced to the Win Score per game average of the team's opponents as compared to the overall NBA average (obviously an above average defense holds its opponents Win Score average below the NBA average). This I consider, again in a slightly dirty way, the team's "defensive" effort. The theory turns out to be 92.6% accurate.

The wins attributed to the offense and defense by the Win Score metric are then, obviously, just "half wins", since each side of the court can only produce 50% of the team's outcomes under my theory. Thus the Celtics have, by my calculation of their opponents' severally below average Win Score per game average, a 72 win defense. But I believe they only have 61 wins because my calculation of their own Win Score per game average reveals a 46 win offense. You can see, however, from that example, that expressing the offensive and defensive win splits as "half wins" makes it easier to understand each unit's strength, because a "full win total" expression, such as "a 72 win season", connotes a certain level of excellence for basketball fans, which they can use to interpret each unit's strength... or to hold the weaker link responsible for its failings.

Here's an example of how. In every season of the George Karl tenure, my calculations say the Bucks were a "50-55 win" offense. Yet they usually posted records that hovered in the .500 range. Why? Because their defense was usually in the low 30 win area. In other words, they had a near championship level offense, and a near lottery level defense, and because of the dichotomy, they produced a slew of mediocre seasons. But, knowing the Win Splits, a Bucks fan can rightfully argue "Had we only had a defense to match our offense, those were all +50 win teams." (And if pigs had wings... but you get the argument I'm making, right?).

Expressing the Win Splits as half wins also allows fans to realistically assess where they stand on each side of the ball. For instance, Bucks fans probably would believe their offense was better than their defense. That is not the case. In fact, the defense, by my measure, was slightly better. But the important point is that both sucked almost equally, and thus a major overhaul of the roster is probably in order.

1 Comments:

At April 10, 2008 at 2:34 PM, Anonymous Anonymous said...

By stating that the Celtics have a "72 win defense" then that requires you to make one of two assumptions -- either they should win 72 games without playing offense at all (which I don't think is what you mean) or that they should win 72 games with their defense and an average offense (offense that would contribute 20.5 wins, or half of a 41-41 season).

Therefore your calculation should be: (average defense)+(average offense)+(Celtics defense +/- average)+(Celtics offense +/- average) = expected wins

If the Celtics have a 72 win defense then it should contribute 51.5 wins, leaving 20.5 to the "average" offense (or, their defense contributes an additional 31 wins).

Similarly, their 46 win offense (which must then assume an average defense) should contribute 25.5 wins (plus 20.5 from an "average" defense).

So your calculation should be: 20.5+20.5+31+5=77 expected wins.

 

Post a Comment

<< Home