La teoria del Dow - michelino di notredame
By: michelino di notredame on Mercoledì 05 Settembre 2001 02:42
C'e' questo articolo del NYTimes di qualche giorno fa. Argomento la teoria di Dow. In estrema sintesi: il mercato gira quando Industrial e Transportation perdono il sincrono. A 11700 dell'Industrial, il Trasportation era indietro. Per cui il turning point era quello. Durata del bear market? Richard Russell, che applica la Dow theory da un mucchio di anni, dice che una previsione di massima si puo' tentare. Il bear market, secondo le regole, dura in genere da 1/4 a 1/2 rispetto al bull market che lo ha preceduto. Insomma, fino al 2003 sicuro. E forse anche un po' di piu'.
A Venerable Market Theory Points to an Extended Slump
By MARK HULBERT
Have some bad news. According to perhaps the oldest market-timing system
still in use, we are firmly in the grip of a long-term bear market.
I refer to the Dow theory, which traces its roots to early last century. Its
creator was William Peter Hamilton, then the editor of The Wall Street
Journal, who developed it in editorials between 1902 and 1929.
Because Mr. Hamilton never codified his approach into a comprehensive
theory, disagreements among followers persist. But there is a consensus on
the overall approach: the trends most likely to persist are those where both
the Dow Jones industrial average and the Dow Jones transportation average
are in alignment. Turning points are signaled when these averages fall out
of sync — events called nonconfirmations.
For example, when the Dow industrials hit its record high of 11,722.98 on
Jan. 14, 2000, the transportation average was more than 20 percent below its
high set in May 1999. Most Dow theorists interpreted that nonconfirmation to
mean that, at a minimum, the bull market was on shaky ground.
But interpretive problems arise. For example, most followers of the Dow
theory agree that the two averages need not hit new highs simultaneously in
order to avoid a nonconfirmation. They disagree about how long we should
wait for the second average to join the first. A week is not enough, but how
about a month or a year? Mr. Hamilton did not provide clear guidance.
Normally, investors should not pay much attention to theories that are this
ambiguous. With so much fuzziness, it is difficult to know whether the
theory itself has a good track record or just the particular interpretation
of it. It is far better to rely on systems that provide clear instructions
on how to react in various circumstances.
An exception can be made for the Dow theory, however, because of a study
three years ago by Stephen J. Brown of New York University, William N.
Goetzmann of Yale and Alok Kumar of Cornell. Instead of trying by themselves
to divine unambiguous rules from Mr. Hamilton's editorials, these
researchers relied on neural networks — artificial intelligence software
that can be "trained" to detect patterns. Upon testing this version of the
Dow theory from Sept. 1, 1930 through Dec. 1, 1997, they found that it beat
the market by a significant margin. (Their study appeared in The Journal of
Finance in August 1998.)
To reach their results, the researchers fed into their neural network the
trading patterns of the two averages in the months before each of the 255
editorials that Mr. Hamilton wrote on this subject. That enabled the network
to learn how to recognize the patterns that preceded a Hamilton buy signal
and the many ways, both subtle and not, in which those patterns differed
from those that preceded a Hamilton sell signal.
The researchers then used the neural network to generate buy and sell
signals for a 67-year period after Mr. Hamilton's death in 1929. For each
day, the researchers fed into their network the market's behavior over the
preceding few months. They assumed that the Dow theory was subject to a buy
signal if stock-price behavior was more like patterns preceding a Hamilton
buy signal. Otherwise, they assumed that a sell signal was in force.
Unfortunately, these researchers have not tested the market's recent trading
patterns. But if they had done so, the system undoubtedly would be issuing
sell signals. Today, the leading Dow theorists of whom I am aware are all
Particularly bearish is the elder statesman among Dow theorists, Richard
Russell, who has been publishing his Dow Theory Letters since 1958. Of the
market-timing strategies monitored by The Hulbert Financial Digest since
1980, his is one of just two that have beaten a buy-and-hold policy adjusted
Mr. Russell points out that the Dow theory does not forecast how long a
trend will persist. He has noted, however, that the typical bear market
lasts one-quarter to half as long as the preceding bull market. That could
mean that this bear market will last at least until 2003.