19-EquityPort Page 573 Friday, March 12, 2004 12:40 PM
Equity Portfolio Management 573According to cognitive psychologists, people tend to overreact to
extreme events. People tend to react more strongly to recent informa-
tion and they tend to heavily discount older information.
The question is, do investors follow the same pattern? That is, do
investors overreact to extreme events? The overreaction hypothesis sug-
gests that when investors react to unanticipated news that will benefit a
company’s stock, the price rise will be greater than it should be given
that information, resulting in a subsequent decline in the price of the
stock. In contrast, the overreaction to unanticipated news that is
expected to adversely affect the economic well-being of a company will
force the price down too much, followed by a subsequent correction
that will increase the price.
If, in fact, the market does overreact, investors may be able to
exploit this to realize positive abnormal returns if they can (1) identify
an extreme event, and (2) determine when the effect of the overreaction
has been impounded in the market price and is ready to reverse. Inves-
tors who are capable of doing this will pursue the following strategies.
When positive news is identified, investors will buy the stock and sell it
before the correction to the overreaction. In the case of negative news,
investors will short the stock and then buy it back to cover the short
position before the correction to the overreaction.Nonlinear Dynamic Models and Chaos
Technical analysis has taken a more scientific twist with the development
of nonlinear dynamics and chaos theory. Patterns generated by nonlinear
dynamic models can be very complex and appear nearly random. A num-
ber of studies have tried to ascertain whether the apparent randomness
of price processes could be generated by deterministic nonlinear pro-
cesses. A chaotic process rapidly becomes unpredictable. There are, how-
ever, chaotic processes that are relatively simple and that maintain a
certain level of predictability. Models of weather, for instance, are cha-
otic but still allow to make reasonable weather forecast.
A number of chaos scientists hoped to discover that economic laws
could be expressed as simple chaotic processes. In particular, it was
hoped to discover that price processes could be described as simple cha-
otic laws with some level of predictability. Should this be the case, chaos
theory offers a reasonable toolbox to recover the chaotic model from
past data. In fact, if the chaotic dynamic is simple, a fundamental theo-
rem of chaos theory, the theorem of Takens, offers a way to fully recon-
struct chaotic dynamics from a sufficient number of past data. In
addition, functional approximation schemes such as neural networks
could be used to approximate the chaotic dynamics.