- I was looking at the offensive ratings over time, and I came up with the 10

worst since 1974:

1 CHI_1999 90.7

2 NYN_1977 91.1

3 NOJ_1975 91.3

4 DEN_2003 91.4

5 PHI_1974 91.8

6 CHI_2000 92.6

7 CHI_1976 92.9

8 NJ_1978 93.6

9 GS_1998 93.9

10 CAP_1974 94.0

Now, that's a pretty crappy list of teams, but somehow it just didn't seem

crappy enough. I thought to myself, well, what really matters isn't how bad

the offense is, but how much the offense is worse than the defense. So,

subtracting offensive rating from defensive rating, I came up with the

following list:

1 DAL_1993 97.6

2 DEN_1998 97.2

3 LAC_2000 96.2

4 VAN_1997 98.6

5 HOU_1983 95.3

6 MIA_1989 96.0

7 LAC_1987 99.6

8 VAN_1996 96.0

9 PHI_1996 100.5

10 CHI_1999 90.7

Again, those teams were bad, but none of those teams (except for Chicago)

had an offensive rating that looks truly horrible. I decided to standardize

the offensive ratings to the league average that year. I got this:

1 DEN_2003 91.4

2 CHI_2000 92.6

3 CHI_1999 90.7

4 LAC_1988 95.6

5 DAL_1993 97.6

6 VAN_1996 96.0

7 DET_1981 96.5

8 NYN_1977 91.1

9 NJ_1978 93.6

10 MIA_1989 96.0

That's a little more like it. Denver's offense, at 91.4 points scored for

every 100 possessions, was nearly 3 standard deviations from the league

average of 102. Seems a little odd, considering the number of exciting

young players, but the Numbers Never Lie.

All of this was just to pass the time. But a more serious question arose: I

only used teams from the 1974 season onward, because that's the point at

which we have complete data. I wondered: has anybody come up with a way to

estimate missing data? Or maybe a simpler question, is there a reliable way

to estimate possessions by using available teams stats? I was going to see

how much error I got using FGA + .4 * FTA, but I wanted to know if anyone's

tried this first so I don't have to reinvent the wheel.

ed - I am not sure exactly what is being computed when you subtract

offensive ratings from defensive rating and get values in the 90s.

Effectively this represents average victory margins per some

standardized possession, no? So shouldn't the range of values

be in the tenths? Whatever.

Regarding the extrapolation of missing stats, in terms of overall

league averages on a year to year basis, it is reasonable (for

getting "close" to the correct answer) to extrapolate rebounding

percentages, as that trend is pretty much non-trending. As for

turnovers, it is pretty clear that this propensity was trending down

over time and, hence, filling in the pre-1974 values for this data

series is largely guesswork.

That said, if one uses what I have argued is the preferred

definition of offensive productivity - what I call "points per

common possession" - in the denominator, which represents

the possession variable, turnovers are ofcourse included and

offensive rebounds are subtracted, and these terms are typically

of equal value. So in terms of a quick and dirty bottom line, using

just shots and reboundable freethrow attempts as an

approximation of total possessions is a decent proxy that gets

worse as one goes back in time from 1974.

*****************

--- In APBR_analysis@yahoogroups.com, igor eduardo küpfer

<igorkupfer@r...> wrote:> I was looking at the offensive ratings over time, and I came up

with the 10

> worst since 1974:

didn't seem

>

> 1 CHI_1999 90.7

> 2 NYN_1977 91.1

> 3 NOJ_1975 91.3

> 4 DEN_2003 91.4

> 5 PHI_1974 91.8

> 6 CHI_2000 92.6

> 7 CHI_1976 92.9

> 8 NJ_1978 93.6

> 9 GS_1998 93.9

> 10 CAP_1974 94.0

>

> Now, that's a pretty crappy list of teams, but somehow it just

> crappy enough. I thought to myself, well, what really matters

isn't how bad

> the offense is, but how much the offense is worse than the

defense. So,

> subtracting offensive rating from defensive rating, I came up

with the

> following list:

for Chicago)

>

> 1 DAL_1993 97.6

> 2 DEN_1998 97.2

> 3 LAC_2000 96.2

> 4 VAN_1997 98.6

> 5 HOU_1983 95.3

> 6 MIA_1989 96.0

> 7 LAC_1987 99.6

> 8 VAN_1996 96.0

> 9 PHI_1996 100.5

> 10 CHI_1999 90.7

>

> Again, those teams were bad, but none of those teams (except

> had an offensive rating that looks truly horrible. I decided to

standardize

> the offensive ratings to the league average that year. I got this:

scored for

>

> 1 DEN_2003 91.4

> 2 CHI_2000 92.6

> 3 CHI_1999 90.7

> 4 LAC_1988 95.6

> 5 DAL_1993 97.6

> 6 VAN_1996 96.0

> 7 DET_1981 96.5

> 8 NYN_1977 91.1

> 9 NJ_1978 93.6

> 10 MIA_1989 96.0

>

> That's a little more like it. Denver's offense, at 91.4 points

> every 100 possessions, was nearly 3 standard deviations from

the league

> average of 102. Seems a little odd, considering the number of

exciting

> young players, but the Numbers Never Lie.

question arose: I

>

> All of this was just to pass the time. But a more serious

> only used teams from the 1974 season onward, because

that's the point at

> which we have complete data. I wondered: has anybody come

up with a way to

> estimate missing data? Or maybe a simpler question, is there

a reliable way

> to estimate possessions by using available teams stats? I

was going to see

> how much error I got using FGA + .4 * FTA, but I wanted to

know if anyone's

> tried this first so I don't have to reinvent the wheel.

>

> ed - Interesting. I did this analysis for my book to some degree actually.

There were a lot of rookie coaches on the bad teams. There was a lot

of player turnover (injuries, etc.). Your list is not that dissimilar

than mine, though I did look at it a few ways, too. There were a

number of other characteristics of the bad teams, too, that I will

save for the book.

As far as filling in the old data -- ain't easy. I have done it

before for the purpose of evaluating Wilt and Russell (also in the

book). The tools to do so just aren't accurate, nor can they be

(fortunately it doesn't make much difference in evaluating Wilt and

Russ). Turnovers are inherently pretty independent of other stats.

Offensive rebounds aren't horrible to estimate, though. But turnovers

make a big difference. The league wide trend only gets you so much.

DeanO

I still owe an update on the sports ec. I should have a chance to

breathe over the weekend.

--- In APBR_analysis@yahoogroups.com, igor eduardo küpfer

<igorkupfer@r...> wrote:> I was looking at the offensive ratings over time, and I came up with

the 10

> worst since 1974:

didn't seem

>

> 1 CHI_1999 90.7

> 2 NYN_1977 91.1

> 3 NOJ_1975 91.3

> 4 DEN_2003 91.4

> 5 PHI_1974 91.8

> 6 CHI_2000 92.6

> 7 CHI_1976 92.9

> 8 NJ_1978 93.6

> 9 GS_1998 93.9

> 10 CAP_1974 94.0

>

> Now, that's a pretty crappy list of teams, but somehow it just

> crappy enough. I thought to myself, well, what really matters isn't

how bad

> the offense is, but how much the offense is worse than the defense. So,

Chicago)

> subtracting offensive rating from defensive rating, I came up with the

> following list:

>

> 1 DAL_1993 97.6

> 2 DEN_1998 97.2

> 3 LAC_2000 96.2

> 4 VAN_1997 98.6

> 5 HOU_1983 95.3

> 6 MIA_1989 96.0

> 7 LAC_1987 99.6

> 8 VAN_1996 96.0

> 9 PHI_1996 100.5

> 10 CHI_1999 90.7

>

> Again, those teams were bad, but none of those teams (except for

> had an offensive rating that looks truly horrible. I decided to

standardize

> the offensive ratings to the league average that year. I got this:

scored for

>

> 1 DEN_2003 91.4

> 2 CHI_2000 92.6

> 3 CHI_1999 90.7

> 4 LAC_1988 95.6

> 5 DAL_1993 97.6

> 6 VAN_1996 96.0

> 7 DET_1981 96.5

> 8 NYN_1977 91.1

> 9 NJ_1978 93.6

> 10 MIA_1989 96.0

>

> That's a little more like it. Denver's offense, at 91.4 points

> every 100 possessions, was nearly 3 standard deviations from the league

arose: I

> average of 102. Seems a little odd, considering the number of exciting

> young players, but the Numbers Never Lie.

>

> All of this was just to pass the time. But a more serious question

> only used teams from the 1974 season onward, because that's the point at

way to

> which we have complete data. I wondered: has anybody come up with a

> estimate missing data? Or maybe a simpler question, is there a

reliable way

> to estimate possessions by using available teams stats? I was going

to see

> how much error I got using FGA + .4 * FTA, but I wanted to know if

anyone's

> tried this first so I don't have to reinvent the wheel.

>

> ed - I have the Nuggets rated as the worst offense of recent history.

Relative to the league average in offensive efficiency, as I measure

it, they are far, far worse than even the 99 Bulls.

--- In APBR_analysis@yahoogroups.com, "schtevie2003" <schtevie@h...>

wrote:> I am not sure exactly what is being computed when you subtract

down

> offensive ratings from defensive rating and get values in the 90s.

> Effectively this represents average victory margins per some

> standardized possession, no? So shouldn't the range of values

> be in the tenths? Whatever.

>

> Regarding the extrapolation of missing stats, in terms of overall

> league averages on a year to year basis, it is reasonable (for

> getting "close" to the correct answer) to extrapolate rebounding

> percentages, as that trend is pretty much non-trending. As for

> turnovers, it is pretty clear that this propensity was trending

> over time and, hence, filling in the pre-1974 values for this data

using

> series is largely guesswork.

>

> That said, if one uses what I have argued is the preferred

> definition of offensive productivity - what I call "points per

> common possession" - in the denominator, which represents

> the possession variable, turnovers are ofcourse included and

> offensive rebounds are subtracted, and these terms are typically

> of equal value. So in terms of a quick and dirty bottom line,

> just shots and reboundable freethrow attempts as an

> approximation of total possessions is a decent proxy that gets

> worse as one goes back in time from 1974.

>

> *****************

>

> --- In APBR_analysis@yahoogroups.com, igor eduardo küpfer

> <igorkupfer@r...> wrote:

> > I was looking at the offensive ratings over time, and I came up

> with the 10

> > worst since 1974:

> >

> > 1 CHI_1999 90.7

> > 2 NYN_1977 91.1

> > 3 NOJ_1975 91.3

> > 4 DEN_2003 91.4

> > 5 PHI_1974 91.8

> > 6 CHI_2000 92.6

> > 7 CHI_1976 92.9

> > 8 NJ_1978 93.6

> > 9 GS_1998 93.9

> > 10 CAP_1974 94.0

> >

> > Now, that's a pretty crappy list of teams, but somehow it just

> didn't seem

> > crappy enough. I thought to myself, well, what really matters

> isn't how bad

> > the offense is, but how much the offense is worse than the

> defense. So,

> > subtracting offensive rating from defensive rating, I came up

> with the

> > following list:

> >

> > 1 DAL_1993 97.6

> > 2 DEN_1998 97.2

> > 3 LAC_2000 96.2

> > 4 VAN_1997 98.6

> > 5 HOU_1983 95.3

> > 6 MIA_1989 96.0

> > 7 LAC_1987 99.6

> > 8 VAN_1996 96.0

> > 9 PHI_1996 100.5

> > 10 CHI_1999 90.7

> >

> > Again, those teams were bad, but none of those teams (except

> for Chicago)

> > had an offensive rating that looks truly horrible. I decided to

> standardize

> > the offensive ratings to the league average that year. I got this:

> >

> > 1 DEN_2003 91.4

> > 2 CHI_2000 92.6

> > 3 CHI_1999 90.7

> > 4 LAC_1988 95.6

> > 5 DAL_1993 97.6

> > 6 VAN_1996 96.0

> > 7 DET_1981 96.5

> > 8 NYN_1977 91.1

> > 9 NJ_1978 93.6

> > 10 MIA_1989 96.0

> >

> > That's a little more like it. Denver's offense, at 91.4 points

> scored for

> > every 100 possessions, was nearly 3 standard deviations from

> the league

> > average of 102. Seems a little odd, considering the number of

> exciting

> > young players, but the Numbers Never Lie.

> >

> > All of this was just to pass the time. But a more serious

> question arose: I

> > only used teams from the 1974 season onward, because

> that's the point at

> > which we have complete data. I wondered: has anybody come

> up with a way to

> > estimate missing data? Or maybe a simpler question, is there

> a reliable way

> > to estimate possessions by using available teams stats? I

> was going to see

> > how much error I got using FGA + .4 * FTA, but I wanted to

> know if anyone's

> > tried this first so I don't have to reinvent the wheel.

> >

> > ed