• ## Re: [APBR_analysis] Re: Assist tracking update

(18)
• NextPrevious
• ... Still preliminary, but that s pretty close to the difference between points per possession and points per play, i.e., the value of an offensive rebound. I
Message 1 of 18 , Feb 3, 2004
View Source
• 0 Attachment
> Try this:
>
>         Shot Attempts   Points  PSA
> AST       412          448    1.09
> UN        555          523    0.94
>
> Value of assist (by my calculations) 0.15

Still preliminary, but that's pretty close to the difference between
points per possession and points per play, i.e., the value of an
offensive rebound.

I have been continuing my conversations with David Berri, the sports
econ professor from Bakersfield.  Using an approach that evaluates
whether wins come from numerous different stats, he keeps seeing that
assists have almost no value.  That approach is different than this
one and I've been trying to make a point that probably can be
clarified just by stats like these (which, as I mentioned before, I
didn't see in the smaller study I did in the early '90's).  I believe
the above is statistically significant.  I'll have to follow up
sometime to find out what games you were doing, though.  You don't
want it to be just a couple teams.

anyone know if dave berri has any basketball related thoughts/musings posted somewhere on the internet, or in print?...

• In the how big is big category, here is how I would contextualize the calculated value. Consider the case of Sacramento, the best passing team in the NBA,
Message 2 of 18 , Feb 3, 2004
View Source
• 0 Attachment
In the "how big is big" category, here is how I would contextualize the calculated
value. Consider the case of Sacramento, the best passing team in the NBA, defeating
its opponents by 7.4 ppg (also perhaps league leading). If we take an assist value of
0.15 and note that they accumulate about 5.5 more assists than the median team in
the league (which by observation appears close to the average) then a fair estimate is
that their exceptional passing ability explains only about one tenth of their
exceptional superiority (calculation: 5.5*0.15/7.4). Does this seem high or low? I
guess that my prior would have been that it would have been a slightly higher.

***********

--- In APBR_analysis@yahoogroups.com, "Dean Oliver" <deano@r...> wrote:
> --- In APBR_analysis@yahoogroups.com, "nick_scholtz" <nick@l...> wrote:
> > Try this:
> >
> > Shot Attempts Points PSA
> > AST 412 448 1.09
> > UN 555 523 0.94
> >
> > Value of assist (by my calculations) 0.15
>
> Still preliminary, but that's pretty close to the difference between
> points per possession and points per play, i.e., the value of an
> offensive rebound.
>
> I have been continuing my conversations with David Berri, the sports
> econ professor from Bakersfield. Using an approach that evaluates
> whether wins come from numerous different stats, he keeps seeing that
> assists have almost no value. That approach is different than this
> one and I've been trying to make a point that probably can be
> clarified just by stats like these (which, as I mentioned before, I
> didn't see in the smaller study I did in the early '90's). I believe
> the above is statistically significant. I'll have to follow up
> sometime to find out what games you were doing, though. You don't
> want it to be just a couple teams.
>
> DeanO
>
> >
> >
> > --- In APBR_analysis@yahoogroups.com, "nick_scholtz" <nick@l...>
> > wrote:
> > > I think the differences in 3pt and FT% obdcure the basic question:
> > >
> > > What is the scoring efficiency on plays with "potential" assists and
> > > on plays with no assists?
> > >
> > > For example, look at the data suplied:
> > > > 2m-2a 2% 3m-3a 3% ftm-fta ft%
> > > > Ast'd 143-263 54% 41-125 33% 39-54 72%
> > > > UnAst'd 200-462 43% 12-43 28% 87-113 77%
> > >
> > > This generated the following.
> > >
> > > Shot attempts Points Scored PSA
> > > Ast 287 366 1.28
> > > Un 512 499 0.98
> > >
> > > Difference in PSA = value of assist = .3
> > >
> > > This is a far lower value than what I would have expected (I
> > > hypothesized a value of 1, John Hollinger uses a value
> > > of .67 for the calculations in the Pro Basketball Prospectus)
> > >
> > > In addition to my suprise at how low the value is, I am suprised by
> > > how few FT's come out of potential assist situations. Perhaps it
> > makes
> > > sense that many fouls come on drives to the basket (not an assist)
> > but
> > > that is a large reason why the calculated value is so low.
> > >
> > > Nick Scholtz
• Plenty of papers available through... http://www.csub.edu/~dberri/ Some abstracts and papers at http://www.csub.edu/%7Edberri/Research.html ...
Message 3 of 18 , Feb 3, 2004
View Source
• 0 Attachment
Plenty of papers available through...

http://www.csub.edu/~dberri/

Some abstracts and papers at

http://www.csub.edu/%7Edberri/Research.html

--- In APBR_analysis@yahoogroups.com, bchaikin@a... wrote:
> > Try this:
> >
> > Shot Attempts Points PSA
> > AST 412 448 1.09
> > UN 555 523 0.94
> >
> > Value of assist (by my calculations) 0.15
>
> Still preliminary, but that's pretty close to the difference between
> points per possession and points per play, i.e., the value of an
> offensive rebound.
>
> I have been continuing my conversations with David Berri, the sports
> econ professor from Bakersfield. Using an approach that evaluates
> whether wins come from numerous different stats, he keeps seeing that
> assists have almost no value. That approach is different than this
> one and I've been trying to make a point that probably can be
> clarified just by stats like these (which, as I mentioned before, I
> didn't see in the smaller study I did in the early '90's). I believe
> the above is statistically significant. I'll have to follow up
> sometime to find out what games you were doing, though. You don't
> want it to be just a couple teams.
>
> anyone know if dave berri has any basketball related
thoughts/musings posted
> somewhere on the internet, or in print?...
• ... defeating ... of ... is ... Perhaps I should also have added the calculation for New Jersey, the second best passing team in the league, as measured by
Message 4 of 18 , Feb 3, 2004
View Source
• 0 Attachment
--- In APBR_analysis@yahoogroups.com, "schtevie2003" <schtevie@h...> wrote:
> In the "how big is big" category, here is how I would contextualize the calculated
> value. Consider the case of Sacramento, the best passing team in the NBA,
defeating
> its opponents by 7.4 ppg (also perhaps league leading). If we take an assist value
of
> 0.15 and note that they accumulate about 5.5 more assists than the median team in
> the league (which by observation appears close to the average) then a fair estimate
is
> that their exceptional passing ability explains only about one tenth of their
> exceptional superiority (calculation: 5.5*0.15/7.4). Does this seem high or low? I
> guess that my prior would have been that it would have been a slightly higher.
>
> ***********
>
Perhaps I should also have added the calculation for New Jersey, the second best
passing team in the league, as measured by assists. For the Nets, assists explain
about 20% of their victory margin (=3.3*0.15/2.4).

The interest in this particular stat is this tentative interpretation: for a rather ordinary
(this year) above-average team, passing, at best (Kidd is the best, no?) can account for
"at most" 20% of success.

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

> --- In APBR_analysis@yahoogroups.com, "Dean Oliver" <deano@r...> wrote:
> > --- In APBR_analysis@yahoogroups.com, "nick_scholtz" <nick@l...> wrote:
> > > Try this:
> > >
> > > Shot Attempts Points PSA
> > > AST 412 448 1.09
> > > UN 555 523 0.94
> > >
> > > Value of assist (by my calculations) 0.15
> >
> > Still preliminary, but that's pretty close to the difference between
> > points per possession and points per play, i.e., the value of an
> > offensive rebound.
> >
> > I have been continuing my conversations with David Berri, the sports
> > econ professor from Bakersfield. Using an approach that evaluates
> > whether wins come from numerous different stats, he keeps seeing that
> > assists have almost no value. That approach is different than this
> > one and I've been trying to make a point that probably can be
> > clarified just by stats like these (which, as I mentioned before, I
> > didn't see in the smaller study I did in the early '90's). I believe
> > the above is statistically significant. I'll have to follow up
> > sometime to find out what games you were doing, though. You don't
> > want it to be just a couple teams.
> >
> > DeanO
> >
> > >
> > >
> > > --- In APBR_analysis@yahoogroups.com, "nick_scholtz" <nick@l...>
> > > wrote:
> > > > I think the differences in 3pt and FT% obdcure the basic question:
> > > >
> > > > What is the scoring efficiency on plays with "potential" assists and
> > > > on plays with no assists?
> > > >
> > > > For example, look at the data suplied:
> > > > > 2m-2a 2% 3m-3a 3% ftm-fta ft%
> > > > > Ast'd 143-263 54% 41-125 33% 39-54 72%
> > > > > UnAst'd 200-462 43% 12-43 28% 87-113 77%
> > > >
> > > > This generated the following.
> > > >
> > > > Shot attempts Points Scored PSA
> > > > Ast 287 366 1.28
> > > > Un 512 499 0.98
> > > >
> > > > Difference in PSA = value of assist = .3
> > > >
> > > > This is a far lower value than what I would have expected (I
> > > > hypothesized a value of 1, John Hollinger uses a value
> > > > of .67 for the calculations in the Pro Basketball Prospectus)
> > > >
> > > > In addition to my suprise at how low the value is, I am suprised by
> > > > how few FT's come out of potential assist situations. Perhaps it
> > > makes
> > > > sense that many fouls come on drives to the basket (not an assist)
> > > but
> > > > that is a large reason why the calculated value is so low.
> > > >
> > > > Nick Scholtz
• So I get to be the first person to correct it :) Talking about calculating the value of assists I offered the ... This was computed from the following data ...
Message 5 of 18 , Feb 5, 2004
View Source
• 0 Attachment
So I get to be the first person to correct it :)

Talking about calculating the value of assists I offered the
following calculation:

> Shot Attempts Points PSA
> AST 412 448 1.09
> UN 555 523 0.94
>
> Value of assist (by my calculations) 0.15

This was computed from the following data

> > 2m-2a 2% 3m-3a 3% ftm-fta ft%
> > Ast'd 143-263 54% 41-125 33% 39-54 72%
> > UnAst'd 200-462 43% 12-43 28% 87-113 77%

The calculation was correct in saying that comparing the assisted
shot attempts non Unassisted the increase in PSA was 0.15 (which
seemed low). This is correct as far as it goes, but I forgot the
final step of distributing the value of the extra points to the
assists. In this case the 412 shot attempts with potential assists
generated 61 more points than if the same shots hadn't had potential
assists. They also were responsible for 184 assists (143 on 2 pt
basktes, 41 on 3 point shots).

This leads to a value of an assists of 61/184 or just under 1/3
point. This is still much lower than I expected but, at least,
should be the correct calculation.

I would be very interested if anyone else has collected similar data
for other teams to see how consistant that number is.

Nick
• After 26 games, here s what it looks like: 2m 2a 2% 3m - 3a 3% ftm - fta ft% AST 311 - 570 55% 97 - 252 38% 97 - 142 68% UNAST
Message 6 of 18 , Feb 12, 2004
View Source
• 0 Attachment
After 26 games, here's what it looks like:

2m 2a 2% 3m - 3a 3% ftm - fta ft%
AST 311 - 570 55% 97 - 252 38% 97 - 142 68%
UNAST 441 - 1037 43% 20 - 95 21% 239 - 318 75%

DIFF 12% 17% -7%
p <.001 <.01 ns

Over the last dozen games I've started tracking shot attempts in the paint
separate from those outside the paint.

In2m - In2a In2% Out2m - Out2a Out2%
AST 89 - 137 65% 69 - 153 45%
UNAST 129 - 275 47% 66 - 192 34%

DIFF 18% 11%
p <.05 ns

We should be able to combine this with the data from the 82games site which
shows shot attempt distance for each player in order to begin to credit
assists properly -- more credit for inside attempts should go to the
assister. By the end of this season, I'll have enough games scored from a
wide variety of teams to know how much of the credit should go to the
passer.

ed
• I ve scored about five Sonics games now ... I need to translate them to HTML. Hopefully the weekend. Are you doing road games as well as home games?
Message 7 of 18 , Feb 13, 2004
View Source
• 0 Attachment
I've scored about five Sonics games now ... I need to translate them
to HTML. Hopefully the weekend.

Are you doing road games as well as home games?
• ... Trying to figure out what to make of these numbers. I ve reduced this chart to one that represents an average game in the sample: .category. 3pt. 2 pt.
Message 8 of 18 , Feb 13, 2004
View Source
• 0 Attachment
--- In APBR_analysis@yahoogroups.com, igor eduardo küpfer
<edkupfer@r...> wrote:
> After 26 games, here's what it looks like:
>
> 2m 2a 2% 3m - 3a 3% ftm - fta ft%
> AST 311 - 570 55% 97 - 252 38% 97 - 142 68%
> UNAST 441 - 1037 43% 20 - 95 21% 239 - 318 75%
>
> DIFF 12% 17% -7%
> p <.001 <.01 ns

Trying to figure out what to make of these numbers.

I've reduced this chart to one that represents an "average" game in
the sample:

.category. 3pt. 2 pt. FT

"assisted" 4-10 12-22 4-6
unassisted 1-5_ 17-40 9-12

Counting 2 FTA as one 'play':

On 35 "assisted" plays, 40 pts are scored, or 1.143 pts per play

On 51 unassisted plays, 46 pts are scored, at .902 ppp

The difference is .216 ppp

But: In reality, only 16 assists are counted. So, from 16 actual,
counted assists, 40 pts are produced.

Therefore, each assist produces 2.5 points; and the difference is
now 1.585 as the value of each assist.

(The actual difference, from Ed's 2184 plays, is 1.561)

We probably all know that a counted assist is only the 'tip of the
iceberg' which represents all 'good passes' a player executes.
While there have been players (Kevin Porter? Ray Williams?) who have
ONLY passed when an assist was to be had, many players get few
assists, yet are credible passers.

So the assist represents an "estimate" of the number of
good, "assist-worthy" passes a player throws.
Your message has been successfully submitted and would be delivered to recipients shortly.