I think the point of the article is that straight statistics like that are not always accurate. For example, what if a player allows the person he is guarding to score more points on average, but during the time is he is on floor, the opposing team tends to score fewer points. This is easily possible if the player is a great help defender who tends to stray from his own man to help his teammates.
This sounds good in theory but there are a TON of variables that would be really hard to encapsulate. I'm not saying it can't be done, but (good) defense in basketball is extremely team-oriented. Positioning, rotations, double teams (and the rotations that accompany them), etc., not to mention zone defense.
To give a more concrete example, a team with a great defensive center gives the guards flexibility to play tighter man to man, because they can more readily afford to let the guy they're guarding get by them (thanks to their "backup", the center). This has the additional consequence of implying that they give up less room on jump shots.
Another thing to consider is steals. A steal is really big because it obviously prevents the opponent from scoring, and frequently leads to fast break opportunities. However, it's misleading to simply look at the number of steals a player gets. Some players get steals within the "flow of the game", whereas others get them through gambling (I don't have a quantitative source for this, but anyone who's watched Larry Hughes play knows what I'm talking about) for steals, meaning that whenever they fail to get the steal, they also end up way out of position, often leading to an easy score for the opponent.
Comments
It seems like you should be able to quantify defense by keeping track of the points scored by the guy you're covering...
Glad to see my hometown team developing new basketball statistics :-)
I think the point of the article is that straight statistics like that are not always accurate. For example, what if a player allows the person he is guarding to score more points on average, but during the time is he is on floor, the opposing team tends to score fewer points. This is easily possible if the player is a great help defender who tends to stray from his own man to help his teammates.
But that doesn't account for who else is on the court with you or against you or for when a team plays zone defense.
Also, help defense makes a huge difference - just ask Marcus Camby :).
> It seems like you should be able to quantify defense by keeping track of the points scored by the guy you're covering...
You'd need to watch deviations from the average points the guy scores. Otherwise whoever is guarding the enemy star player looks bad.
This sounds good in theory but there are a TON of variables that would be really hard to encapsulate. I'm not saying it can't be done, but (good) defense in basketball is extremely team-oriented. Positioning, rotations, double teams (and the rotations that accompany them), etc., not to mention zone defense.
To give a more concrete example, a team with a great defensive center gives the guards flexibility to play tighter man to man, because they can more readily afford to let the guy they're guarding get by them (thanks to their "backup", the center). This has the additional consequence of implying that they give up less room on jump shots.
Another thing to consider is steals. A steal is really big because it obviously prevents the opponent from scoring, and frequently leads to fast break opportunities. However, it's misleading to simply look at the number of steals a player gets. Some players get steals within the "flow of the game", whereas others get them through gambling (I don't have a quantitative source for this, but anyone who's watched Larry Hughes play knows what I'm talking about) for steals, meaning that whenever they fail to get the steal, they also end up way out of position, often leading to an easy score for the opponent.
The list of variables goes on and on....