- Here is my take on the numbers provided - that is on the big picture of the Allen/Payton swap. To assess overall offensive and defensive performance (as I haveMessage 1 of 34 , Apr 9, 2003View SourceHere is my take on the numbers provided - that is on the big picture of
the Allen/Payton swap.
To assess overall offensive and defensive performance (as I have argued
in the contentious string on intertemporal game simulation) it is
highly preferred to look at "big possessions" - what I call common
possessions (CP) - which document the number of times per game the ball
changes hands. This definition provides a consistent basis for
comparing offenses and defenses. (The particular formula used is
defined in a posting in the aforementioned string.)
That said, here is how I see the numbers going.
SONICS: (Before = 52 games, After = 24 games)
Before: CP/game = 92.47, P/CP = 0.995, opp.P/CP = 1.008
After: CP/game = 90.60, P/CP = 1.040, opp.P/CP = 0.986
Change: CP/game = -1.865, P/CP = 0.045, opp.P/CP = -0.022
Bottom line: a "slowing" of the game of 2%, a rather dramatic overall
improvement of 0.067 points per common possession, two thirds of which
is offense and one third defense.
Also, relative to the NBA average performance of about 1 P/CP (a guess
equal to last years average, from which there likely has been little
change) the offense has gone from a below average offensive and
defensive team to above average in both categories.
The CAVEAT is that there may be significant discrepancies in terms of
the home and away schedules and the quality of opponents between the
BUCKS: (Before = 54 games, After = 24 games)
Before: CP/game = 92.8, P/CP = 1.055, opp.P/CP = 1.057
After: CP/game = 94.8, P/CP = 1.079, opp.P/CP = 1.081
Change: CP/game = 2.0, P/CP = 0.024, opp.P/CP = 0.024
Bottom line: an increase in game pace of 2.2%, and offsetting
improvement and worsening of offense and defense, respectively.
And the Bucks results are somewhat robust as the winning percentage of
opponents is essentially 50% before and after (not sure exactly what
the definition of the statistic was - end of sample, or when game
played). The qualification "somewhat" above is noted as I did not check
whether there was a serious discrepancy in home versus away games in
the Bucks data.
- ... This is actually misleading. I won t say wrong, but misleading. Pace is very important. Many of the individual stats are offensive and without adjustmentMessage 34 of 34 , Apr 11, 2003View Source--- In APBR_analysis@yahoogroups.com, "Mike G" <msg_53@h...> wrote:
> What matters is that x points and y rebounds occurred in z minutes.This is actually misleading. I won't say wrong, but misleading.
Pace is very important. Many of the individual stats are offensive
and without adjustment for pace overweight players who just put up a
lot of offensive stats _per minute_. And pace is mostly (not
entirely) a function of a decision about pace, not ability (though
ability has a small impact). How fast or slow teams play reflects
primarily on their style not their substance. How many points and
rebounds per minute reflects both style and substance. Removing pace
more isolates the substance.
The way common possessions do it is by isolating another variable in
addition to minutes that is common between both teams on the floor.
Both teams have 48 minutes to win. Both teams have 100 possessions
to win in a fast game or 85 in a slow game or whatever. But both
opponents have the same number (not true with small possessions
= "plays"). So, yes, what matters is what happens per minute, but it
is equivalent to say that what matters is what happens per
possession. And what happens per possession isolates quality better
than what happens per minute.