Well, what performance measure is for, if you don't think it helps you win games? If you think some statistical measure is good performance and help you win games, if you think who and who on the floor together help you win games, you have an underlying "model". Oh, LOLN in practice is a model, because the theorem is only true asymptotically. PS. Res: Make them really understand, not just remember some theorem on the book.
Teams do not play consistently. There always extraneous factors, or you could say randomness. The goal is to make the best, most informed prediction of how two teams will play in a given situation. The more information or knowledge you have concerning the past, the better equipped you are to do that. You're right, it's important to ask the right questions. The meaningful questions. That's kind of the point of this thread. I invited posters, such as yourself, to ask some meaningful questions that could potentially be answered by looking at the play by plays, and which could tell us something interesting about the team. The numbers in the first post, as I stated, were just a quick, starting example. If there's something else that you'd find more interesting, then ask away. If you're saying there is nothing interesting that can be gleaned from the play by plays, then this thread isn't for you.
I'd like to know Yao Ming's post play efficiency difference between when he wait to see the defensive setup, and when he make a move right away. I don't know if they track a stat like that.
I'm certain the Rockets track it, but that's not something that can be determined from the play by plays. It's a very good question, though, and I think JVG and Adelman had different philosophies. Adelman is always preaching patience, while as I remember it JVG wanted Yao to make quick, definitive moves when he got the ball. Again, if you or anyone else has any questions that can be answered, or at least decently approximated, using the play by plays, please ask and I can try to figure it out. That's my intention with this thread.
Why do you say that defense can't be measured? Obviously, it's much harder to measure it, and the play by plays aren't sufficient, but it is measurable. As for chemistry, I'll say the same thing I said about leadership. If it translates in an observable way on the court, is it not in some sense measurable?
I'd say it's quantifiable but not measurable in any useful manner. You can say it has a positive affect and you can point to some evidence of that but the concept itself is vague enough to have so many variables that a linear measurement is simply not sufficient.
If its quantifiable (and therefore can be represented as a "stat'), and it can be determined that it has some positive effect, then I think that it is useful information to track. Down the line, when people get smarter and more of those variables are recoded, maybe someone figures out a way to plug in the chemistry information into a grand model.
Probably cant be tracked, but I'd like to know Ron Artest's 3pt % on the catch n shoot vs off the dribble. Guess assisted 3 pointer is as close as we can get. Like to know McGrady's as well, because he looked like a different(much better) shooter off the pass than off the dribble.
Unfortunately, even percentage on assisted 3-pointers is difficult to determine, because there's no way to distinguish between an assisted missed shot and an unassisted missed shot. Potential assists aren't tracked in the play by plays.
The pbp is the problem. What people call "stats" is too restrictive. This is why data warehouses and data mining are huge industries - get the most amount of data you can. Get everything. Then see if you can come up with something quantitative from that. Getting the information first is the problem. I once had this grand idea to come up with a scheme to track possessions using some shorthand I'd make up that would track the movement of the ball, who touched it, and upon "the outcome" (a shot, a turnover, whatever...) would note things like position on the floor the ball was shot from, was the guy being guarded or was he open on the shot, was it rushed, how much time was left on the clock, etc. Once the game was done, each posession would basically be a row of text that could be parsed by a program and broken down into stats that could be dumped into a SQL database and queried at will. But then I realized I didn't have the time to do it. Others on this board have come up with their own ways to scout/score a game and have posted it in the past. Have you ever read Dean Oliver's "Basketball on Paper" book? That's some fun reading if you're into quantifying basketball using stats and want to go beyond ppg, rebs, etc.
Thought this little bit of info may be appreciated in the BBS... Speaking of statistics, data guru and our current GM is now on Bill Simmons latest podcast over on espn. Im not a member and can not post new threads but wanted to get this info out. For those unfamiliar with Bill Simmons I find him fairly entertaining and his podcast with Morey is pretty good.
Here's the podcast: http://podloc.andohs.net/dloadTrack...gin/mp3/espnradio/sportsguy/simmons090309.mp3
I've heard countless people tell me this is an awesome read but I kept forgetting to order it for whatever reason. Just did now. Thanks for the reminder.
If you haven't already seen this, here's an old website of his while you wait for the book : Journal of Basketball Studies
Basketball on Paper is a great book. I read it a few years back. He uses a short hand charting notation in the book like you describe, but I never tried it out for myself.
