P.S. I should've added a qualification: none of this
implies that there aren't some players and even teams out
there that may be shooting too many long 2-pointers.
But we can't simply look at their FG% and conclude that
they should take fewer 2-pters and more 3pters. Just
as we can't look at Aaron McKie's 46% shooting and
conclude that he should shoot more and Iverson at
37% should shoot less. Maybe they should, and maybe
they shouldn't. The raw FG%s are not sufficient data
from which to draw a conclusion.
Oh, another example of how a low FG% shot may be the
one to take: the stereotypical example is when the
team's PG is stuck with the ball and 2 seconds are
left on the shot clock. Certain players, usually
PGs, are the ones who get stuck with the lousy
shots. The player may not have wanted to put up
a shot with a 30% chance of success, but that shot
may nonetheless have been the best option, compared
to a 24-second violation.
From: Michael Tamada
Sent: Friday, August 13, 2004 3:44 PM
Subject: 2pters vs 3pters (was: [APBR_analysis] Re: Who is the next
From: Dean Oliver [mailto:deano@...
Sent: Friday, August 13, 2004 9:51 AM
>> > you're just watching one the sideline: Guys who take lots of long
>> > 2-point shots will KILL you. Show me a guy who does this and I can
>> And this from the guy who bemoaned the disappearance of the mid-
>> range shot?
>> I know there's a difference between midrange and a long two, but I
>> think it's subtle at best. For shame, John, for shame. :)
>a foot on the line. What makes it hard is that where a shot should be
>taken is a very individual-specific thing. If you're making 40% of
>your midrange j's, you're about breaking even. Roland has identified
>several guys who do this. Can't complain about those guys taking the
>shot. But the league average of 35% or so means it is, on average, a
The other complication is that even shots which are on average bad can
still be an important and useful shot to take. Sometimes one has
to, and even wants to, take a lower percentage shot.
Paradoxical? Here's three examples which resolve the paradox.
1. NBA teams probably average about a 95% FG% on dunks. That's a
lot higher than any other shot. Therefore, team that want offensive
success should do nothing but dunk.
Clearly preposterous, teams have to shoot some other shots. They
might prefer a 95% dunk attempt, but maybe the 35% or 40% 2-pointer
is in fact the best shot available.
2. More realistic example: if teams shoot ... rather than looking
it up, I'll guess ... 60% on inside shots, and 36% on 3-pointers
(yielding an equivalent FG% of about 54%), and only 35% or 40% on
long 2-pointers, they nonetheless might still WANT to shoot some
of those 2-pointers. Because if they don't, their offense becomes
too predictable and easy to defend -- and eventually their
scoring efficiency would decline. Teams need to have diverse,
It's the opposite extreme of the example that DeanO gave of bad
coaching in the last 10 seconds of the game: give the ball to
your best player and have him create something. If the defense
knows that's what you're going to do, their defense becomes much
In contrast, if your team has several offensive options (not
just choice of players, choice of shots that they might take),
the defense's job is harder.
3. From football: unless they've got Ryan Leaf at QB, practically
every football team averages more yards per pass attempt (including
incomplete passes) than they do yards per rushing attempt. So why
don't teams do nothing but pass and stop running those inefficient
Again, predictablility vs diversity. Your quarterback will get
mangled and your receivers double-teamed if the defense knows
that you will do nothing but pass. It's easy to show with a
simple game theory matrix that even though the "rush" option
has a lower average payoff than the "pass" option, teams want
to randomly throw a few rushing plays in there, just to keep
the defense off balance.
(There's also the issue of the spread or standard deviation
that one gets from rushing vs passing plays, but I'm ignoring
that in this example.)
Oh, a 4th example, from finance. Suppose Microsoft stock has
an expected return of 5%, with a standard deviation of 4%.
Meanwhile, MBI stock has an expected return of 4%, also with
a standard deviation of 4%. What would your optimal
portfolio look like?
The knee jerk reaction is to buy only Microsoft and avoid
that dog MBI stock. But again, it's easy to show that the
optimal portfolio, assuming one has any risk aversion
(desire for low standard deviation) at all, is to buy at
least a little MBI stock (unless its price movements are
perfectly correlated with Microsoft's). As any wise
financial advisor will tell you: diversify, diversify,
Despite the lower average return, an optimal portfolio needs
to have a little MBI stock in it. And despite the lower effective
FG%, an optimal offense needs to have some long 2-pointers in it.
Yahoo! Groups Links