Re: When to leave a game?

Expand Messages
• ... on the article, but ... Will do. Hopefully tonight, but life/work may get in the way. Generally, they assumed that the percentage of possessions on which
Message 1 of 6 , May 6, 2003
• 0 Attachment
--- In APBR_analysis@yahoogroups.com, <igorkupfer@r...> wrote:
>
>
on the article, but
> the math was slightly beyond me.

Will do. Hopefully tonight, but life/work may get in the way.

Generally, they assumed that the percentage of possessions on which a
team scored was optimal at some pace and lower at either a higher or
lower pace. Then they simulated the games and, sure enough, the team
won fewer games at higher and lower paces than at their optimal
pace. My point has been that we don't know where that optimal pace
is and that the optimal pace in basketball may be actually a large
range of paces. The average time of a possession is about 15 s and I
believe that teams perform about equally well whether they use 10 s
or 20 s (beyond about 20 s, rather, with <=3s on the clock, they do
worse).

DeanO
• ... something ... on the article, but ... p. 43: We remark that for simplicity we will think of a team s unconstrained mode as its best scoring offensive
Message 2 of 6 , May 7, 2003
• 0 Attachment
--- In APBR_analysis@yahoogroups.com, <igorkupfer@r...> wrote:
> > I wasn't very impressed with that article on pace management,
> > though. They pretty much assumed the answer that they got,
something
> > I discussed with schtevie offline for a while.
> >
> > Not sure if there are any others.
> >
> > DeanO
>
>
on the article, but
> the math was slightly beyond me.

p. 43: "We remark that for simplicity we will think of a team's
unconstrained mode as its best scoring offensive style of play versus
the current opponent."

This is the assumption that ruins everything. They assume that there
is exactly one pace at which the teams is optimal, rather than an
array of paces or a range of paces.

p. 44: "A popular belief among sports analysts is that, when the
score is tied or a team is ahead, a lengthy possession is better than
a short possessions. Surprisingly this belief, as a general
statement is false."

This is the kind of analysis that is nice to do, but they biased
their answer by assuming that a team gets worse by holding the ball.
Now that may be true if they hold it too long -- I have seen that --
but they need more to strengthen such a statement.

Figures 1 through 3 use the notation (P, time units), where P is the
probability of scoring and time units is how many time units each
possession takes. Note that in each case, the intermediate time unit
case (=2) has the highest P. If they had left all the P the same and
just changed the pace, their arguments would have been potentially
stronger, but now their results are clouded by changing P.