<div class="quote_poster">Quote:</div><div class="quote_post">As I mentioned when I first released the stat I have developed called Defensive Composite Score, it was very much a work in progress. I?m not one to assume that my ideas are better than others, so, after listening to e-mails from readers and ideas from the good people at the APBRmetrics message board (see http://sonicscentral.com/apbrmetrics/index.php), I made a few changes to the statistic. In my opinion, it?s better than it?s ever been and it passes the ?laugh test? better than ever before (although the importance of that is debatable). The biggest change that was made was getting rid of the method of ranking the players. While this was a simple way of doing things, it did not make it clear how close or far apart some players were. Player A was better than Player B but worse than Player C, and that?s all you knew. But by how much? Instead, the new DCS is based on z-scores (thanks to reader Dave Elkin for the idea). The only downside to this is that the ratings are not on a simple 0-100 scale. However, it?s still very possible to understand what a player?s rating truly means (more on this later). Another major change was the removal of the box score stats completely. The three stats used were blocks, steals, and foul efficiency. Unfortunately, those stats don?t tell us much and can be very deceiving. I replaced the box score section with the counterpart PER statistics that can be found here at 82games. Because counterpart PER is generally not equal for all positions, I calculated it based on a player?s z-score within their position. Also, I consider small forwards and shooting guards to be interchangeable, so those two positions were combined. Finally, I received a bunch of e-mails explaining how it?s not fair for Player X to get a certain score because said player either plays only in garbage time or said player has to guard the other team?s best scorer every night. I tried to adjust for that by using a statistic developed at the APBRmetrics board called %St. This statistic estimates how much of a player?s minutes were against starter-quality opponents. Now, the top of the leaderboard isn?t dominated by players who come in off the bench to guard lower quality opponents. The new average for DCS is 0. Positive scores denote good defenders and negative scores mean bad defenders. </div> http://www.82games.com/nichols2.htm Here are the league's top 10 defenders according to 82games: Tim Duncan 73.0 Joel Przybilla 61.5 Bruce Bowen 58.4 Yao Ming 57.0 Manu Ginobili 56.3 Kevin Garnett 54.5 Zydrunas Ilgauskas 49.9 Tyrus Thomas 48.5 Jermaine O'Neal 48.0 Devin Harris 47.6
This was posted in the Nets forum a few days ago, and I still stand by what I said then - "82games.com is usually great with statistics, but these are just plain inaccurate. I'm not talking about the way they were calculated but rather the results. Yao Ming a better defender than Shane Battier? Rudy Gay and Pau Gasol good defenders? There's a lot of weird stuff like that in this study."
<div class="quote_poster">Voodoo Child Wrote</div><div class="quote_post">This was posted in the Nets forum a few days ago, and I still stand by what I said then - "82games.com is usually great with statistics, but these are just plain inaccurate. I'm not talking about the way they were calculated but rather the results. Yao Ming a better defender than Shane Battier? Rudy Gay and Pau Gasol good defenders? There's a lot of weird stuff like that in this study."</div> This stat seems to be more of a defensive impact on a game though. Obviously these don't gauge a person's individual one-on-one defense but guys like Yao or Pau have more of an impact on defense being shotblockers, space eaters in the paint, rebounders, etc. They affect the other team's FG%, what shots they take, protect the rim and deter players from the paint. Battier is obviously the better individual defensive player but his impact is limited to the man hes defending and when he goes to take a charge or help out on someone else's man.
<div class="quote_poster">Voodoo Child Wrote</div><div class="quote_post">This was posted in the Nets forum a few days ago, and I still stand by what I said then - "82games.com is usually great with statistics, but these are just plain inaccurate. I'm not talking about the way they were calculated but rather the results. Yao Ming a better defender than Shane Battier? Rudy Gay and Pau Gasol good defenders? There's a lot of weird stuff like that in this study."</div> Memphis was such a horrible defensive team last year that Gay and Gasol end up looking like above average defenders because the Grizzlies defense wasn't as bad with them on the floor (still pretty bad, though). They're basically benefiting from the defensive ineptitude of the entire roster. And Battier's score was lower than it probably should be because they 82games.com's counterpart PER data isn't accurate for players who crossmatch on defense. Battier was generally assigned to the best perimeter offensive player, not simply the opposing SF. Conversely, Yao benefitted from a low counterpart PER (most centers simply aren't offensive threats). Yao's individual defense probably ends up looking better than it actually is.
I wonder how much time it took them to put that together. It seems at least somewhat accurate, but as always, there is really no objective way to rate defensive players.
Honestly I don't really look into these in depth stats like this. I think complicated statistics like these can mislead you.
