Devising a New Formula for Offensive Value

[Shinyhubcaps]

Hello everybody,

I've been struggling through a season of watching the Milwaukee Bucks play a terrible style of offense, and I know that something has to give. Too many times, especially lately, I've seen Earl Boykins and Drew Gooden NOT pass the ball, hoist up a terrible shot, and shoot a low percentage. Somehow, I want to quantify these things (lack of ball movement, poor shot selection, poor shooting) into one number for offensive value.

Now I expect that a similar number may already exist, though I do not know how the things are weighted or what is included.

Anyway, here is my first attempt at an Offensive Value formula:

http://img51.imageshack.us/img51/5546/o ... evalue.png

http://www.mediafire.com/?ir2p63o1ta9yu6r

Unfortunately, I've only been able to include the Bucks so far, but I may plug in known players from other teams for comparison's sake. For now, though, I need an objective statement on a few things.

1.) Are all items appropriately selected? (Meaning, are these good indicators for a smooth-flowing offense versus a clogged drain of futility like we have in Milwaukee.)

2.) Does the formula make sense? (Meaning, are correct, like items compared, and positive items and negative items used appropriately such that a high shooting percentage results in a better rating, etc.)

3.) In what ways would I be able to normalize the data? (Meaning, 100 is average for the league, for example. Maggette is one of the worst shooters against the league average, and that should be reflected perhaps with a rating lower than 100.)

4.) Is this a useful tool, and what else could I add? (Meaning, Brockman and Mbah a Moute are not offensively skilled, but they know their place in the offense enough such as to not disrupt that. However, is it understood that they are not go-to guys, and that going to them more would not improve the offense necessarily?)

Thank you to everyone who looks this over and anyone who has any suggestions for amendments. I can also answer any questions that you may have in the meantime.

EDIT: I'll add any players here that I run after-the-fact for comparison from other teams.

Steve Nash: 130.56

[ElGee]

Could you explain the formula more and the thinking behind it? It will help answer the other questions. I assume you know Oliver's formula...

[Shinyhubcaps]

The formula is meant to identify players who perhaps are failing to "let the offense come to them." It should identify when players take too many shots at a low clip ("chuckers"), as well as players who bottleneck the pace of the offense (fewer team assists with them on the court, as well as lower team output with them on the court).

The difference from Dean Oliver's Offensive Rating is that he uses Points Produced, which as I understand it, accounts for individual assists and points, but not a "+/-" style on-court/off-court split of team output.

It's also intended to be simpler, as it does not (yet at least) incorporate turnovers or rebounds as relevant to how a player works within the team offense. I would say it also doesn't require team offensive rating as an input, whereas Oliver's rating utilizes team offensive rating to normalize the data, but the team offense is relevant in the on-court and off-court splits.

[Shinyhubcaps]

Shoot, I just realized that the way the parts are combined doesn't make sense.

W is the multiplier which rewards players for being on the floor when a higher % of FG are assisted. Conversely, players who play in lineups that do not move the ball are penalized.

Example: 6% fewer FGs are assisted with Earl Boykins on the court. His multiplier is 100% - 6% = 94%.

X, Y, and Z though, don't make sense to add because X = ~100, Y = ~0.5, and Z = ~0.5 for several players. Thus, X (essentially points per 100 possessions) is too heavily weighted dealing with these magnitudes.

[Darain]

Does this formula make Lebron look better than Kobe?

[Shinyhubcaps]

Does this formula make Lebron look better than Kobe?

That's not the intent behind it. I'm a knowledgeable NBA fan and care about many teams and players, not someone in piddly LeBron v. Kobe debates.

But since you mention it:

LeBron James = 116.02 (a bit too little ball movement with him on the court)
Kobe Bryant = 118.94 (big difference in team scoring with him on the court)

Although interestingly enough, Oliver's Offensive Rating has James ahead of Bryant 113-112.