PDA

View Full Version : Project Defensive Stat Scoresheet



mepanchin
05-13-2007, 03:04 PM
I recently read a book referenced on Pomeroy's site called Basketball on Paper by a guy named Dean Oliver. Mr. Oliver is a very bright engineer who develops astute statistical models of estimating the offensive and defensive performance of individual players. I'm currently in the process of applying his formulas to Duke teams going back to 1987 (when goduke.com's stats site stops listing detailed individual stats), but I wanted to suggest that others take up the cause with me and, for this coming season, record the relevant defensive statistics (since the relevant offensive ones are easy to come by).

The relevant statistics are:
- Forced field goal misses (FM) - recorded as each time a particular player's man misses a shot. Blocks are added to these to create total forced field goal misses. When an unguarded man misses a shot, it is a team FM. When a help defender comes over to help cause an offensive player to miss, then both the main defender and the help defender get half a FM.

- Forced turnovers (FTO) - When a defensive player forces an offensive player to commit a turnover. This is separated into Steals and Forced turnovers and combine into Total Forced Turnovers.

- Forced free throw misses (FFTA) - when a defender fouls an offensive player who then misses free throws.

- Allowed field goals (DFGM) - When a defender's man scores on him.

- Allowed free throws (DFTM) - When a defender fouls a man who then makes free throws.

Add fouls, blocks, steals and defensive rebounds and you have a fairly robust defensive statsheet that can be used to (pretty accurately) derive a good rating and stop percentage for individual players.

So, for this upcoming season, I am going to do my best to record these stats for each Duke game, but more people should join me so we make sure we are keeping the most accurate stats we can. This will really give us a good picture of how well our players are defending without just using our perception which can betray us.

Anyone interested?

cspan37421
05-13-2007, 08:32 PM
JMO but FTO sounds really subjective and the validity of including FFTA is really questionable. If a defender fouled Billy King in the act of shooting should they really be seen as a better defender than someone who fouled JJ Redick? Isn't really just that they were lucky in getting assigned a below-avg FT shooter? And it gets really hairy when you're playing something other than man-to-man.

Well good luck with it anyway.

mepanchin
05-13-2007, 09:06 PM
To respond:

1. FTO is not any more subjective than assists. Every scorekeeper will record assists differently. For instance, some will consider it an assist if you pass to someone who has a fairly open drive to the basket but needs to take several dribbles while another won't. Scorekeepers' discretion is already a big part of basketball stat sheets, no reason to consider it a big deal now. Furthermore, the fact that discretion plays a part makes it even more useful to have multiple people interested in it.

2. Seems to me that a defender should know the opposition well enough to know when to foul and when not to. Good defending means doing everything possible to keep the other team from scoring, including fouling a bad free throw shooter hard when he tries close shots, and restraining against guys like Redick who obviously shoot well.

A defensive rating is going to, by essence, be a function of the team's ability to defend as much as it is individual prowess, but that is no different than an individual's ability to perform offensively on one team vis-a-vis another. I can list numerous players who played better (offensively) on good offensive teams and worse on bad offensive teams.

As is, I have finished calculations approximating individual defensive and offensive efficiency for 89-94 and 97-07 Duke teams. The offensive ratings are far more accurate than the defensive ratings, but they should at least be consistent and fairly accurate. The math is pretty complicated, but it makes sense and is a reasonable way of approximating defensive and offensive prowess. However, if I am able to have a defensive scoresheet for every Duke game next year, I can more closely calculate defensive efficiency (it's not worth it to go play-by-play offensively because the ratings formula is really quite accurate).

cspan37421
05-13-2007, 09:46 PM
Good points, esp. about assists. Well, still seems like a latta work. ;-) Maybe you can bounce your ideas off of Wojo whoever does the team's number crunching. I'm sure K has a way of measuring defensive prowess.

mepanchin
05-13-2007, 10:07 PM
Oh I'm sure they do. As is, from general team stats, blocks, fouls and steals, one can approximate defensive efficiency for each player and it seems to be pretty accurate. For instance, last year our best defensive players (according to my calculations) were McClure, McRoberts and Nelson, which validates what we all saw subjectively. Henderson was fairly average defensively, and Scheyer and Paulus were fairly bad defensively. But, recording defensive stats like mentioned above would make the calculations more accurate.

At any rate, having others interested in it would be nice as well. I think all it would really take is a cursory viewing of each game twice. Once normally (since I get too emotionally riled up to keep score during the game anyway) and once more or less in fast forward in order to find what you're looking for (FMs, FTOs etc). I guess it's a bit of an investment, but others must be stat nerds like me!