Sharechat Logo

Forum Archive Index - February 2003

Please note usage of the Forum is subject to the Terms & Conditions.

Messages by Date [ Next by Date Previous by Date ]
Messages by Thread [ Next by Thread Previous by Thread ]
Post to the Forum [ New message Reply to this message ]
Printable version

Re: [sharechat] Diversification Theory

From: Travis Morien <>
Date: Tue, 25 Feb 2003 12:25:23 -0800 (PST)

--- Capitalist <> wrote:

> There's another flaw with the diversification
> theory, beyond not accounting
> the value of your time (as Steve pointed out) and
> related to the asymmetric
> information problem (as noted by Kirez): the
> assumption that
> correlative/non-correlative relationships are
> invariant and/or linear over
> time.

My goodness, so many big words.  Why use a big word
when a diminutive one would do?  ;)

I don't think anyone does assume that correlations are
always the same.

The theory behind modern portfolio theory is just that
the volatility of a portfolio consisting of two or
more assets whose correlation is not a perfect one
will be less than the mean volatility of the
individual investments.  the measure of the
diversification merit of an investment is the

> The only way that the diversification of a portfolio
> can reduce risk (by
> smoothing out deviations from expected value) is
> when the relationships
> between the price movements of each of the
> investments contained within the
> diversified portfolio are truly random

No.  This isn't true.  As I just said, as long as they
are not perfectly correlated with a correlation
exactly equal to one there will be some
diversification merit in terms of reduction in

> (non-correlative, either positively
> or negatively) or corellating in a known, invariant
> pattern.  If the market
> behavior of the various elements of the diversified
> portfolio starts to
> correlate in any significant way, or the
> corellations start to fluctuate
> significantly, I would argue that risk exposure is
> actually magnified
> disproportionately over a non-diversified investment
> strategy.

Constructive interferance huh?  Like when waves
interact and you get a doubled sound, or beats?


If all the assets suddenly performed the same
(correlation = unity) then the portfolio would have
the same volatility as any one asset.

Remember, we aren't putting 100% into each asset,
we're putting a fraction.

What the poster seems to be claiming is that if you
get a portfolio of five assets that go up, or down, by
10% that these would add.  You'd lose 50% of the

Can someone explain to me how that maths works again?

A $100,000 portfolio has 5 assets of $20,000 each. 
They all suddenly perform exactly the same and lose

I say the loss would be 5 x ($20,000 x 10%) = $10,000

But the post I'm replying to says that the losses
would in fact exceed $10,000.

I'd love for someone to show me how that works because
I did four years of maths, chemistry and engineering
at uni and I can't for the life of me see how the loss
would exceed $10,000.
> Since true randomness is a theoretical ideal, and
> not a true reality
> (especially when we are talking about the highly
> complex
> inter-relationships
> which spring from economic activity),
> diversification can only support a
> portfolio during those times when the markets'
> behavior most closely
> approximates the ideal of randomness or corellative
> invariance.  The
> problem
> is, the markets' behavior doesn't always do that. 
> It is nearly impossible
> to know when the correlative relationships are all
> going to change out of
> their established patterns, which they do with some
> regularity.

One could argue that one should not build portfolios
based exclusively on computerised historical
optimisation and expect that this will necessarily
lead to an efficient portfolio, in fact one does. 
Mean variance optimisers are referred to by wags as
"error maximisers" for just this reason.

> A lot of diversified portfolios got hammered over
> the last few years simply
> because the relationships between investment sectors
> were not actually
> random, and started showing some new heavy positive
> correlation (to the

It had nothing to do with randomness or non
randomness, it had to do with if you had a large
equities weighting, especially US and Europe, then you
probably lost money because those sectors lost money.

Your property investments, cash, bonds and mortgages
would by and large be ok but if you were two thirds in
equities you'd have lost.  Its common sense and nobody
ever said diversified portfolio never lose money in
any year.

> downside).  If the majority of market sectors are
> down, diversification
> isn't going to help you at all (and might actually
> hurt you).

Oh really?  You just figured that out all by yourself?

> Frankly, diversification theory appears to me to be
> a rationalization
> contrived to provide psychological support for
> investor complacence, and I
> think this is related to the point Kirez was trying
> to make about Buffet,
> the value of researching your investments in depth,
> and building value.
> Diversification theory says, essentially, that if we
> go by a few general
> assumptions about the markets, we can invest our
> money without having to do
> a lot of research into the particulars of each
> investment.  In order to
> offset the increased risk inherent to putting your
> money in investments you
> don't really know in depth, we are supposed to
> spread the money around so
> that any bad bet won't ruin the whole portfolio. 
> The problem then arises
> as
> to what happens when entire market sectors, or even
> sectors of sectors,
> start moving together against you.  That's when you
> discover that
> diversified market theory merely transfers your
> gamble from a select few
> investments onto broader economic forces (which are
> far less easy to
> analyse
> and understand, and which you really haven't tried
> to understand beyond a
> statistical approximation in formulating your
> diversified investment plan).

Such a long paragraph, where to begin except to point
out once again that yes indeed if the market goes down
a market portfolio will also go down.  It doesn't
require a genius to note this.

If the argument here is that if everyone did intense
research nobody would lose money then this is flatly
impossible.  Somebody has to lose money because the
average return of every investor will be the same as
the market return before costs.  It is a mathematical
impossibility for everyone to make money unless the
whole market makes money.

Now the market is a highly competitive place.  Every
single market participant is trying hard to make money
and for the most part working reasonably hard to score
an advantage.  Clearly only an exceptional individual
is going to be able to outperform by a large extent
over the long term.

But if your name isn't Buffett or Munger you should
diversify.  Funnily enough diversification is very
much recommended by other investment luminaries such
as John Templeton and Peter Lynch who were known to
own huge numbers of stocks.  (more so Templeton, who
also owned bonds, property and whatever else he
thought looked cheap).
> This then brings up an epistemological problem that
> relates directly to the
> nature of statistical reasoning itself (since
> diversified portfolio theory
> relies so heavily on statistical reasoning). 
> Statistical reasoning is an
> epistemic method for evaluating the relative truth
> of competing claims in
> the absense of definitive particular information in
> support or
> contradiction

Good Lord your thesaurus is going to be sore tonight. 
A basic premise of good writing, even in technical
writing one should try to keep it clear.  This passage
has the same ring to it as a great deal of modern
philosophy where essentially it has no meaning and
little logic but the words are all done in such a way
that few readers can follow it but its all very clever

There was a famous scandal a number of years ago where
a prankster actually deliberately wrote a passage
absolutely devoid of meaning but used terms so complex
and "smart" sounding that few dared challenge what was
under it all just a load of crap.  It was accepted for
publishing by a leading journal of modern philosophy
and following that our friend went to the newspapers
to point out that one of the world's most prestigious
philosophy journals had accepted and published what
was in fact a deliberately random bunch of words. 
Caused quite a stink among the arty farty types.  8P

> of any one of them.  In investing, the information
> is largely available (as
> Warren Buffet would tell you) to allow you to form a
> judgment without
> relying on statistics for approximations (you can
> get all the financial
> statements for Company A and Company B, along with
> supporting information,
> to evaluate their relative strength as investment
> prospects).  Using
> diversification as a primary investment method is
> effectively the choice to
> *ignore* available specific information in favor of
> approximations (e.g.
> assuming that the particular information is NOT
> available, even when it
> usually is).

If this information is so easily available and easily
interpreted then why is the market not completely
efficient then?  Many people adopt the diversified
portfolio approach, many don't.  There are still under
and overvalued stocks so it clearly isn't as simple as
this writer implies.

I agree that focus investing is a valid strategy if
you've got the time and the ability, but diversified
portfolio are another approach and provided you don't
unrealistically expect strong positive growth in every
single year, even years when many world markets are
down 20-30%, you should do well in the long term.


Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, more

To remove yourself from this list, please use the form at


Messages by Date [ Next by Date: Re: [sharechat] Housing Travis Morien
Previous by Date: [sharechat] Housing michael gore ]
Messages by Thread [ Next by Thread: [sharechat] PWC prhughes
Previous by Thread: Re: [sharechat] Diversification Theory ]
Post to the Forum [ New message Reply to this message ]