Lamrock wrote:I'm pretty sure Bayless will be forced into the rotation somehow, Blake and Miller's minutes will be closer to even and Webster will either be dealt or play garbage time. Might be wrong though. Looking forward to the next set of updates.
Well, I'm not a fan of anything so I don't stay up on all the hot gossip, I was just eyeballing the minutes based on how much they played recently, where are they likely to play (i.e. with the Blazers without Miller example, because Bayless is so bad, I assumed they'd be willing to let Roy handle the ball primarily and play Fernandez alongside him), what's an ideal rotation, etc. I'll gladly take suggestions for minutes to put the guys at, or even wild fantasies.
Benji I'm assuming you live in the Northeast cause of the thunderstorm. Anyways my questions is how are you coming up with these stats? Like what references and sites are you using?
Michigan. We get them, they power up over the Great Lakes and as they pass over the Frozen Wasteland then you guys get them.
Anyway, shadowgrin is mostly right. From B-R.com I only get the totals, I used to get them from nba.com, but B-R's "CSV" makes copy/pasty super easy. All the stats on there are easy enough to calculate and have been in my files for years now. Then I go to 82games.com to acquire the data from the play-by-play. See:
http://www.82games.com/0809/08DET11.HTM for an example. Everything is extracted out of this base data. For example for the three second shooting above, 82games has the individual value, it takes a couple calculations and you can see the players share of the team.
Some like Jae and KevC have seen the files.
Here's an older example for the Bucks that has about 90% of what the final database of this year has. This is about 15% bigger than the one from 2007-08, and so on. The older ones have been scrapped, I'll have to re-enter the data at some point back to 2002-03 because it's grown a ton since 2007. I'm constantly adding things, adding columns to "try things out" and so forth. The file now requires the 2007 Excel format, the old xls is too small and can't handle the number of columns. There's some stuff which should just be trimmed out from the main file anyway. I have a raw data file (not organized into teams, etc.) for sorting I could throw up if there's interest. I usually separate it by positions (C/PF, SF/SG, PG) and combine the players who got traded, but I have yet to do that.
Anyway, for the projections, basically it works like this. A teams possessions are made up by player possessions. We can define the player possessions, then add it up for a team, or distribute a teams possessions based on that info. In this case, we're doing both. Essentially we say, if these players play together, and there's this many total possessions, and they play this many minutes, we can expect this usage, and knowing how efficient they are we can extrapolate this point production. Then you add it all up and get the team values. Then it's a simple pythag to get the record off the point differential.
The concept of finite possessions is very important, and often overlooked. A player like a Joel Pryzbilla or Erick Dampier is more valuable to a team with stars than Jermaine O'Neal or Antoine Walker because they don't use possessions, when they do it's very efficient (i.e. they mostly only dunk or tip in shots) and they also get offensive boards saving possessions. O'Neal and Walker use a ton of possessions (taking them from teammates which would be okay except...) they're very inefficient in them so they kill the team even while getting gaudy per game numbers. If you have Walker or O'Neal and they pull up for a three pointer or a jumper out of the high post, that's a shot their teammates can't have. That's not to say these players can't be valuable (i.e. like offloading all the bad possessions to them or bailing the team out...or say if O'Neal became an offensive rebounder, star defender who didn't need ten bad jumpers a game, or as Walker redefined himself for the 2006 Heat title as a bail out ball handler or spot up shooter) but we can't really ever know when a player will actually adjust how they play so we have to assume they'll perform as they did in the past.
This year I've been looking at using 2007-08 or 2008-09 data to make projections instead of just the prior season. Especially in cases of outlier years. (Redd and Bogut on the Bucks this year were one if we assume they play as well as they did in the 30 or so games this year, we should expect the Bucks to be a 45-50 win team with them playing 82 games along with Sessions, Mbah a Moute, random PF all starting. But if you use the 2007-08 data which is closer to their norms, we're looking at 40-ish wins best case.) Of course, we can never know when a player has a fluke year (Murray having a career year off the Hawks bench after looking like he'd be out of the league in the prior two years) or rejuvenation (Shaq deciding to wreck everyone last year) not to mention injuries. In many cases the projections are "best case" scenarios such as last years Pacers looking like a 50 win team, then losing Dunleavy for almost the entire year among others. (And coaches who never find the obvious optimal rotation. Or teams, like the Wizards, who insist on playing DeShawn Stevenson, Juan Dixon, Mike James, etc.)
BTW: If anyone is interested in the 2008-09 projections (which were not as advanced as the ones this year can be if I get around to them) they are in
this post. Always remember, these were based on the year prior, no injuries, optimal rotations, etc. Lulz at Pistons and Hawks. I can't forsee major trades like Miller/Salmons, or say LeBron having a historic season or Durant finding his star potential, either. That said, I've of course learned from last year, which is why I'm taking far more things into account this time around.
That said,
I did warn you about the Clippers. (And I had them fully healthy!) (
Also Ronny Turiaf.)
Blazers at 5.... FAIL
Post of that thread.
Did this simple:
- Code: Select all
Player MPG
wade,dwyane 38.9
o'neal,jermaine 36.5
chalmers,mario 34.1
beasley,michael 34.1
haslem,udonis 34.1
cook,daequan 17.0
diawara,yakhouba 17.0
quinn,chris 14.6
anthony,joel 7.3
jones,james 4.9
magloire,jamaal 4.9
41-41
Odom replacing Haslem: 43-39
0809Boozer replacing Haslem: 35-47
0708Boozer replacing Haslem: 42-40
