I found all the current statistics misleading in certain ways. You have to have good basketball understanding and a group of statistics to draw some fairly useful conclusions. For example, more steals maybe a result of gambling on defense with many out-of-position scenario's not shown on stats sheet. More points maybe a result of fast paced offense with volumn shooting attempts. Good penetration leads to open shots for teammates, but the teammate just can't shoot(no assist recorded). Not to mention hussle, box out, save an out-of-bound ball... Why can't we just create a more intuitive statistics to evaluate players? For instance, Brooks used a good screen of scola and scored a 3-pter--Brooks +3 and scola +1(or +0.5?). At the same time, if nash didn't even try to fight through the screen, Nash -3. If he did fight through and forced Brooks taking a contested difficult shot and it went in, Nash 0. If scola get called moving screen when Nash fighting through, Scola -1 and Nash +1. If the moving screen is called because Brooks didn't wait enough for scola to set the pick, then Brooks -1, scola 0, Nash +1... This can be applied to charges, box out, put a hand in the face of the shooter, 3 seconds, win the jump ball, good penetration, draw fouls... We can +0.5, +1, +1.5... for all kinds of good basketball performance, no matter if it's offense, defense, intangible or clutch performance. We then punish all the bad performance even if it's not on stats sheet at all. We can then calculate the weighted summary on per minute, per 48 minutes basis. It's difficult to track on a play by play basis, but once it's tracked it'll be enormously intuitive and useful. Some subjective judgement maybe involved, and we may assign less weight to such scenario's. I'd call the final statistics 'true impact statistics'.
I agree there are a lot of interesting things we'd like to track that don't show up in the boxscore, or even the play-by-play. And, for sure, the Rockets have a host of people that are doing just this during the game. Its hard to say how many points a particular play is worth. Its pretty subjective. Sometimes, a team can do all the right things, and the shot just don't go down. Or on another possession, a team can do all the wrong things, but they score by a fluke. What's the proper way to grade such actions?
It's not that easy, otherwise it would exist. However, I have been meaning to propose an idea for a while: How about making up a Google Docs spreadsheet for clutchfans guys to collect game data? It wouldn't have boxscore details, but other numbers e.g. fighting through picks, # of screens set, # of transition 3pters to name a few. I don't know what 'irregular' statistics are important, but it would be pretty useful to log any and every event on the court. Organizing something like this will obviously be very difficult, but if we take assign people to certain entries, and have our own rotation system, I think it would be a very fun thing to do. And people wouldn't have to commit too much time to study something like this. As fans of a team that has a good statistical minded GM, I think we should be inspired to try and have something like that. I know Durvasa is one of the better known promoters of statistics on here, but one person can only do so much. Together though, we might be able to to tackle something like this with small steps from everyone. I know I need time to think up a decent system that would work with the number of people on here, but I'll start doing that now. It's a group project so any ideas would of course be welcome.
There could be many ways to tackle this, but I personally prefer a system rewarding 'good basketball performance' regardless of outcome. So a good penetration leading to wide open shot for teammate will +1 by itself, and a awkward pass which leads to teammate hitting that difficult shot will be 0, but +2(or +3 if it's 3 pter) for that teammate. My thoery is that over time the result will be in sync with the performance, so the 'flukes' will be evened out over time. But I agree it's kind of subjective and may need some adjustments.
"Fighting through picks"? Some of those are judgmental calls... that referees don't even call. Instead of those judgment calls, why not keep stats on the stuff that CAN be measured and that actually does get a call from the official: Charges taken. 3-point plays "The Old Fashioned Way." (fouled and shot went in, shows strength to finish a play when fouled), Fouls in offensive end of the court (committed a silly foul when got the ball stolen, and fouled the guy before fast break, although it may seem like a good thing, though), Fast break assists (shows unselfishness), or Fast break shots made (shows ability to finish the fast break wisely) Those are just some examples of things that can be measured without wondering "did he really fight through the pick?" or "did he really set a screen?". You have the right idea, though. Please write proper grammar and punctuation. It makes it easier to read, sir.
It's subjective and leads to interpretation. But theoretically it can be done. Here are some stats that I think would do a much better job of painting a picture of how well a player does at certain aspects of the game: jump-shot percentage (outside the paint) fouls drawn from inside the paint picks set successfully picks utilized successfully average time of possession after crossing mid court FG% based on zone I would think that such analysis could/should be performed per quarter and also during the last 5 minutes of the game. I would also think that it would be a good idea to markup those numbers compared to which defender a play is playing against that night. Stats like these could be used to answer such questions as: does he hold the ball a long time? where, on the floor is he most comfortable shooting? How competitive is his defense? What kind of plays work into his favor? Is he a slasher? a shooter? a driver? a dunker? Is is fast? powerful? both?
That's where the method needs adjustments. I'll put less weight on these situations. But everythings else you mentioned should be counted in.
