The Halloween
Indicator
In Beating the Dow O’Higgins argues that 85
percent of gains with the DJIA (just capital gains, not dividends) occur
between October 31 and April 30. He claims to have deduced this observation from
data between 1925 and 1989. He points out that several market analysts have
established that certain months have historically been significantly better for
the market than others for a variety of reasons. His own explanation for the
Halloween Indicator is fragmentary and presents a different argument for each
month.
Whatever the explanation, it is worth
noting that hard facts corroborate this view. For example, both the 1929 and
the 1987 crash occurred during the non-fertile summer semester. Furthermore,
despite the fact that the 1929 crash signaled the beginning of a many-year
decline, immediately following the crash the market increased—if modestly—until
April 1930.
In An S-Shaped Trail to Wall Street I confirmed O’Higgins’ claim. Over the 30-year history
that I examined in detail, the DJIA gains during winter semesters were the only gains. With two more years of
history we can now update these observations and draw conclusions about where
we are heading.
Exhibits 3 and 4 show the detailed evolution of
capital returns from the DJIA over the last 32 years. We see both
semester-by-semester and 5-year averages separately for the winter and the
summer seasons. Between October 1968 and October 2000, winter semesters were
characterized on the average by a 9.9% gain, whereas summers by a 0.7% loss.
Exhibit 3 shows a gentle but steady rise of the
5-year average gains. Such a rise is much less noticeable in the summer
semesters (Exhibit 4). During the last two years the average gains differed by
14.4 percentage points (13.3% gain in the winters compared to 1.1% loss in the
summers). Such a difference seems far from the usual 10 percentage points. We
may expect smaller differences between winters and summers in the next few
years.
Exhibit 3. The evolution of capital
returns from the DOW during winter semesters, defined as October 31 to April
30, during the last 32 years. The 5-year average (purple line) shows a steady
slow rise.
Exhibit 4. The evolution of capital returns from the DOW during summer semesters,
defined as April 30 to October 31, during the last 32 years. The 5-year average
(purple line) shows a not-very-convincing rise.
The fact that the phenomenon of winter-summer
difference persists undaunted indicates that investors have not become aware
of, or simply have not reacted in an important way to this piece of
information. Otherwise the Efficient Market Hypothesis* would have wiped the phenomenon out. But to the extent that it still
exists, it constitutes an opportunity. Risk-averse strategies can indeed
benefit from portfolios with six-months-on six-months-off the market
strategies.