Bucks Diary

Tuesday, November 18, 2008

I am launching a new NBA power rankings blog

I'll be doing limited Bucks basketball blogging for a day or two. All of my free time is being invested in the launch of my new basketball blog which I am calling "NBA Advanced Power Rankings".

The new blog is gonna be pretty sweet, and I promise that once I get it up it won't take away from my Bucks writing in any way. In fact I believe it will enhance the NBA experience for Bucks fans and NBA fans in general.

The new blog will provide an objective, continuously updated, NBA Power Ranking list based entirely upon my "Point Value over Average" relative performance grading system. The ranking list will be updated every morning... or right after I calculate the PVOA's for the last night's games and then compute the required averages.

The blog will also provide a PVOA analysis of each individual NBA game, going back to the games played on November 11, 2008. The analysis the score of the game, the estimated number of possessions in the game, each team's overall Point Value over Average total based on the possession estimates, and each team's offensive and defensive PVOAs.

Finally, the blog will include a weekly "Trend Report" which will provide fans with the percentage of below average performances (Overall, Offensive, and Defensive) by each NBA team in that particular week.

Initially "NBA Advanced Power Rankings" will simply be a crude source of the above referenced information. Eventually I want it to be a high functioning cousin to similar football blogs like FootballOutsiders.com (the inspiration for the PVOA system) and AdvancedNFLStats.com. In other words, a reliable source point for objective information about the relative strength of each team in the National Basketball Association.

My NBA Manhattan Project

I also have one other project in the works, which I'm calling my "NBA Manhattan Project". I am going to try to build a database of seasonal Win Profiles for every NBA team dating back to 1960.

I never thought I could do Win Profiles for any season prior to 2001 because of the lack of necessary statistical information. But two things happened that made me believe it could be done.

First, I realized my Offensive/Defensive Win Score system (the system I use to generate "Win Profiles") hinges entirely upon each player's Marginal Win Score. That means I can calculate any given player's Win Totals for any given year without knowing the NBA positional averages in that given year.

But I still didn't have the statistical information I needed, such as: a. each player's counterpart opponents' statistics and, b. essential Win Score statistics the NBA didn't keep before 1976-77.

I figured out how to overcome those limitations with the help of baseball essayist Bill James. In his "Historical Abstract" he makes the argument that Jackie Robinson was underrated. As part of his argument, he cites Robinson's excellent "zone rating" at second base during the 1948 season.

"Zone Rating" is an ultra-modern baseball defensive statistic that estimates each fielders defensive performance according to the number of balls hit into the fielder's "zone". Making such an estimate requires information that was not gathered until recently. So how did James know Robinson's 1948 rating?

Reading further, I figured it out. Bill James estimated Robinson's zone rating using existing information to deduce the required but unknown information. (Specifically, he found a strong relationship between the number of "assists" recorded by a team and the number of groundballs.)

That's when it hit me. I could do the same thing with Win Profiles. I can estimate each player's defensive statistics by simply adjusting modern statistical averages and then projecting them backward. Furthermore, I can estimate each team's Win Score and Opponent Win Score by establishing consistent relationships between the available statistics and all of the unknown statistics.

Obviously, that means everything before 1977 will involve two layers of estimates. That won't be a problem, though. I did mock Win Profiles for last season using only the information available in 1960. I then compared the results to my actual Win Profiles from last season. The two sets of numbers had a 90% correlation.

So I decided to do a full NBA Win Profile abstract database, which I will complete over the course of the next year. Once it is complete I intend to use the information produced to write a book that will examine historical debates (was Chamberlain actually better than Russell? Did Magic Johnson ever have a season equivalent to Oscar Robertson's famous "Triple Double" season? ) and will compare players across eras and across positions. It will be similar in nature to Bill James baseball abstract. Stay tuned.


At November 18, 2008 at 7:57 PM, Anonymous Anonymous said...

You are a badass of solar proportions.
Way to go!!

At November 18, 2008 at 9:33 PM, Anonymous Blake said...

Wow you are gonna be very busy.

I already put your new blog into my favorites, can't wait to see it!!

At November 21, 2008 at 3:24 PM, Anonymous Anonymous said...

Nice going man, I'm glad you will be expanding your work out to a larger audience. I think it will pay off for you.



Post a Comment

Links to this post:

Create a Link

<< Home