Fooled by Randomness is a bit
difficult to review, because, as the author himself observes, it is
more like a series of essays. As a result, the point of the book
seems to wander. However, a number of themes can be observed.
Perhaps the main point is that what we often attribute to skill is
really due to randomness. Sometimes, as with dentists or
pianists, skill is required for success. However, in business,
and particularly in finance, one can be successful simply because one's
business or trading temperament match the conditions of the
market. Given the large number of traders, the probability of one
of them being consistently successful over a multiple year period due
to pure luck is very high. However, if their success is based on
luck, sooner or later luck will do them in. As a long-term market
trader, Taleb has gone through a large number of colleagues.
Inevitably, the (many) stories of his colleagues end with the phrase
"they blew up," meaning that they lost substantially more money over
the course of a few days (sometimes sooner) than the total sum of the
money they had made over the course of the previous years.
Invariably, Taleb gives two reasons for the blow-ups. The first
is that the traders thought their success was due to their system,
whether it was an economic approach or an algorithmic approach;
in reality, they just happened to be in the right place at the right
time with the right skills and temperament. The second is that
they failed to prepare for the case in which their system failed,
because they did not believe their system could fail. If the
trader lacked the brains to understand what they were doing, then they
tended to blow up because of too much leverage (due to overconfidence
in their supposed skills). On the other hand, if the trader
trusted too much in brains, they followed their philosophy when the
market conditions no longer matched the assumptions of the
philosophy. By contrast, the few successful traders were less
successful in the good times, but they constantly protected
themselves. In fact, Taleb even specializes in making money off
the rare, so-called "black swan" events: he tends to lose a
little bit on most of his trades but makes a large amount on a few
trades.
Taleb observes that most of our training in probability is with
symmetrical distributions like the normal distribution. In these,
the downside is usually equal (but opposite) to the upside.
Often, however, this is not the case. In finance, in particular,
the upside may be small but the downside very large (or
vice-versa). So instead of looking at the probability of
something happening, it is better to look at the expectation value,
which will include the asymmetry. This idea of handling "black
swan" events in one's system of trading is one of the main themes.
Another theme is that people are not designed to correctly deal with
probability: even the experts in probability make glaring
errors. It appears that our emotions interfere with the
rationality required for correct probabilistic thinking. These
are necessary to allow us to decide between equally valued
options: if the philosopher's donkey had been randomly nearer
either the hay or the water, the deadlock would have been broken.
But in finance, we may have to trick ourselves, for instance, by not
looking at our results very frequently, since losses are emotionally
worse than the emotional high from an equal gain.
Our propensity for incorrect probabilistic thinking manifests itself in
a few common ways. One major way is through the fallacy of
thinking that repeated observations say something meaningful about the
probability distribution. Repeated observations will typically
not encounter the rare events, so if these are of substantial
importance, observation does not yield useful information.
Likewise, if the probability distribution is changing (as it does in
finance, where everyone else in the market is constantly adjusting
their strategy to exploit current anomalies), the previously collected
information is no longer relevant.
Another major cause of incorrect thinking is due to success-bias.
Often we attempt to derive causes for success from the successful
population (the successes did X). This is meaningless unless we
also include the failure population. The fact that the successful
traders did X is not meaningful if there were a large number of
failures who also did X. Likewise, if a medical diagnosis is 95%
correct, the probability of having the disease if diagnosed with it is
not 95%, it is 1/(50 incorrect + 1 correct) = 1/51 ~= 2%.
A final theme of the book is the author's contempt for non-intellectual
thinking. This is leveled most frequently against journalists,
whom he claims are purveying entertainment but think they are providing
information. However anyone without an emphasis on brains gets a
dose: MBAs, overconfident financial traders, and the general
public. While he has some good points, the arrogance practically
drips off the page. I have held similar attitudes in the past
(and to some extent, still in the present), but this attitude does not
promote healthy relationships with other people. In fact, I get
the feeling that, given enough time, the author will withdraw from the
world, kind of like the citizens of C.S. Lewis' fictional Hell in The Great Divorce, who are so full
of themselves that they keep moving away from the other citizens
because they cannot stand their company.
Fooled By Randomness is easy
to read, due to the conversational style. Unfortunately, the lack
of strong structure inhibits a context for the ideas, and the excessive
parenthetical comments are rather distracting, as are the interesting
but tangential ideas. However,
the ideas are expressed simply but adequately in a way that is easy to
comprehend. While this is not a 100 year book, it is a good
primer in practical statistics, and the repeated lessons about not
"blowing up" a caution to future investors.
Review: 7
The ideas are very good, but really
need to be presented in a nicer way. The arrogance, while
understandable, is tiring. The parenthetical comments are excessive and
distracting; I have found that if I am tempted to put more than
one parenthetical comment in a paragraph, I am well served to remove or
rephrase it. Finally, the ideas are excellent, and could be
worked into a 100 year book with some actual discipline, such as the
editors whose advice he ignores.
Copyright © 2010 by Geoffrey Prewett