Burns on Cramer

John,

We definitely agree on one thing: Cramer's goals appear
to be having fun and getting fame.  I have no way of knowing
if he is having fun, but he has certainly succeeded with fame.
(Disclosure: Two of my top goals are having fun and not getting
famous.  At present I'm doing well with both these.)

I think goals are of secondary rather than primary importance.
Reading my horoscope in the newspaper may satisfy my goal
of knowing how to run my life, but I think a more important
question is whether that is the action of a rational person or not.

What I am advocating is SCIENCE: What can we know? And
then let's set about learning that.

What we can learn with random portfolios is whether or not a
fund or a market commentator exhibits skill over some period of
time.  (There are other things to learn with random portfolios as
well, by the way.)

Once we have information on skill (which as far as I know is
uniquely available through random portfolios), then we can address
the issue of goals.  We can get a sense of how much skill was needed
to meet the goal over the past period -- maybe a lot, maybe even
negative skill would have done.  More importantly we get a hint of
how we might best set about to achieve our goals in the future.

If we focus on goals rather than skill, we have very little basis on
which to carry past performance into the future.

If you wanted to pick a poker player to back, would you merely
look at the player's winnings or would you try to evaluate if he or
she outperformed given the hands that were dealt?

In terms of what is contracted for, I have previously suggested that a
good choice is to contract for exhibited skill as determined by random
portfolios.  This rewards fund managers that do exhibit skill, and allows
the investor a mechanism that comes closer to paying for what they get.

Pat

Patrick Burns
patrick at burns-stat.com
+44 (0)20 8525 0696
http://www.burns-stat.com

I read this paper,
http://www.burns-stat.com/pages/Working/cramer_vs_pseudocramer.pdf,
and found it to be interesting, however, the idea of judging
real-world results against artificially constructed portfolios leaves
me cold. The only reasonable way of judging performance is against
stated goals. Goals tend to be specified in terms of returns (relative
or absolute),  variability (volatility, draw down and so forth) or a
combination (Sharpe, Sortino...) and the only reasonable question is
to what extent the goals are met. Judgment against a basket of random
portfolios, however cleverly constructed, is simply not defined in
relation to the efforts of the manager.

(In this particular case, one might well ask what Cramer's goals are.
They would seem to be to have fun and gather fame. Since it seems that
he is eminently successful on both counts one is forced to acknowledge
that he is doing a good job.)

As for the challenge to chartists in the paper's conclusion, they too
should be measured individually against their goals, not random
portfolios. Why this emphasis on goals? Because this goals are what
investors pay for. They may use past performance as a gage to
ascertain whether the goal is obtainable, but by and large investors
pay for specified goals and retain or fire managers on whether those
goals are met. Other assessment alternatives matter little, even if
they are 'better', unless investors agree and contract for them. This
is after all a contractual relationship and results need to be
evaluated in terms of expectations.

   jab