## Fwd: Re: 0.4 vs 0.48 (Hoopstudies)

Expand Messages
• Message 1 of 1 , Feb 16, 2002
>This came from "jamisont1"

>From: "jamisont1" <jamisont1@...>
>To: "HoopStudies" <deano@...>
>Subject: Re: 0.4 vs 0.48 (Hoopstudies)
>
>
> >>>>Basically, the estimate formula using 0.4 ends up an average of
>about 2% too high (91 poss vs 89 poss actual).
>
>You are right Dean, +2 more possessions per game than actual
>possessions if you use 0.4.
>
>I think John and Dean both are Correct on this one.
>Using 0.48 is more right if you only consider the free throw
>percentage only, and using 0.35 is more right if u consider total
>possession per game.
>
>The actual percentage of technicals, bonus free throws, and 3point
>fouls is a lil higher than 0.4 (im not sure how much higher but I
>calculated alleyoops2's free thow data and it was 0.43%)
>
>but if you put 0.48 on this (Possessions=FGA-OR+TO+0.4*FTA),
>possession error margin will be larger.
>
>Why eror margins is larger for using more accurate number?
>cuz that formular has a big error on blocked shots.
>
>lets say.. you shoot a ball (shooting attempt) then get blocked, it
>goes out of bounds or ur teammates picks it up. then you still keep
>the ball.
>
>real possession is just 1 since there was no change of possession,
>but its 2 possessions according to that formular, +1 error margin)
>
>I looked at NBA team's average block shots per game. it was 5.4 last
>year.
>When you get blocked, i think theres more than 50% chance your team
>will get it back. (alot of block shots just go out of bounds)
>
>so there is more than +2.2 error on that formular per game
>consistantly.
>thats why using 0.35 or 0.32 is more accurate than using 0.48
>
>Am i right on this? I hope i do....
>damn its 4 am.
>
>
>--- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
> > --- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
> > > I'm also seeing that more flagrant fouls and more technicals due
>to
> > > the new defensive 3-s rule might actually be lowering the 0.4. I
> > did
> > > the Det-NJN game, the Den-Mem game, and the Sac-Was game from
> > > yesterday. In all cases multipliers of 0.32-0.35 work best. My
> > > original work back in the mid '80's started with a multiplier of
> > > 0.36, I remember. But that was based on the 1987 NBA Finals
> > alone.
> > > That plus a little follow-up was good enough to last 15 years, I
> > > guess.
> >
> > I'm not so sure I really trust all the online play-by-play stuff.
> > There are definitely errors, ones that I thought were minor, but
>may
> > not be. I did sum things up for all 2/14 games and found the best
> > multiplier for all games. They ranged from 0.22-0.35 with an
>average
> > of 0.30 and a std dev of 0.05. At the end of a lot of play-by-
>plays,
> > there were funky things happening and occasional gaffes in the
>middle
> > of them. If anything, these would appear to lower my counts of
> > possessions.
> >
> > Basically, the estimate formula using 0.4 ends up an average of