Timing the Market

Patterns in American Stock Market Movements

by Ben Best




Timing the Market refers to having foreknowledge of movements in the price of a security, index or derivation prior to the occurance of that movement. Clearly, if someone can "time the market" that person can benefit by buying or selling prior to the anticipated price movement. It is often said that "you can't time the market", yet there is ample evidence that certain price movements can be anticipated correctly more often than not.

In the "PUBLISHED OBSERVATIONS" section I have collected published information from a number of sources, including THE WALL STREET JOURNAL, BARRON'S, BUSINESS WEEK, MONEY and other publications. I have also drawn information from Stock Trader's Almanac, Ned Davis Research and Stock Market History. I sometimes give citations and sometimes do not -- let the reader beware! For a broader understanding of investing & trading strategies -- plus more market timing information -- see my article, Investing & Trading in Equities: Art & Science.

In the "MY OBSERVATIONS" section I provide results of my own computer-analysis of four major American stock market indices -- the DOW, the NASDAQ-100, the S&P 500 and the Russell 2000. For background concerning these indices -- see my article The Major American Equity Indices.

(Historical note: Bull Market refers to an upward-moving stock market, whereas Bear Market refers to a downward-moving stock market. The symbols were first used in the London stock market in the 1710s based on the idea that an attacking bull thrusts upward with its head, whereas a bear fights using downward movements of its paws).

(return to contents)


The strategy "sell in May and go away" when subjected to data analysis yields some persuasive evidence for the possibility of "timing the market". According to Ned Davis Research $10,000 invested in the DOW 30 (DJIA) in 1950 for the November 1st to April 30th period each year would have yielded $415,890 by 2001, in contrast to $1,743 in the May to October period. The profitable period is extended to include May & October during bull markets, whereas in bear markets November & April can suffer seriously. The effect may be stronger for technology stocks -- an estimated 99% of the total technology return was earned in the mid-October to mid-March period in the 17 years preceeding 2002.

If equities are not held in the June to September period, money should not lay idle, but should be invested in fixed-income securities. (For information on fixed-income investing, see my article Fixed-Income Securities: Money-Markets & Bonds).

In the 1971-2001 period, January & April were the best months for the DOW 30 (2.3% each), whereas September & August were the worst (-1% & 0%). For the NASDAQ Composite, January & December were the best months (4.30% & 2.50%), while October & September were the worst (-0.60% & -0.20%). The last five trading days and the first two trading days of a year typically display a "Santa Claus rally". From 1959 to 2001 the S&P 500 averaged a 1.5% return in this period.

The fact that January, April and July were good months for the DOW 30 in the last 50 years of the 20th century may have been associated with being the first month of the quarter. October, although it begins the fourth quarter, may suffer from being so close to September. This downside risk is most often attributed to mutual fund portfolio adjustments, which are typically completed by the end of October.

October is perhaps the most volatile month. Much of October's reputation comes from big crashes, most notably in 1929 & 1987. Removing the large losses for a few bad years makes October one of the best months for the DOW. October is the turn-around month, when bear markets become bull markets -- although the opposite happened in 1929.

March is similarly a month of dramatic change -- "in like a lion, out like a lamb" should be "in like a bull, out like a bear". Stocks typically sell-off at the end of March, possibly because money is being taken-out of stock and being sent to the Internal Revenue Service prior to April 15th.

In the year 1900 agricultural activity made August the best month of the year for the stock market -- and August continued to be the best month until 1952. But in the 1990s August became the worst month. The so-called "summer rally" is probably an historical artifact of the agrarian economy. In the last 38 years of the 20th century, DOW 30 "rallies" (quarterly lows to quarterly highs) were 13.7% in Winter, 10.9% in Spring, 10.3% in Autumn and 9.6% in Summer. Economic activity associated with the Christmas season and inflows to retirement funds are an explanation given for the Winter/Spring advantage.

Between September 2, 1997 and July 24, 2001 the DOW 30 index rose 2618.70 points. During that period there were 47 months with 47 first-trading-days which saw a 2662.51 point gain. During that same 47 months there were 933 non-first-trading-days for which there was a 43.81 point loss. Possibly this pattern occurs because so many company retirement funds mechanically invest on the first trading day of the month -- or because many employees buy stocks with their discretionary funds when paid at month-end.

Another pattern in the American stock market is the effect of Presidential politics. Using the DOW index from 1886 and Cowles & other indices to 1832, the final two years of a President's term showed nearly 3 times the total stock market gain as the first two years of office. Since 1937 the average stock-price move between the lowest point in the second Presidential year and the highest point in the third year has been 50%. Wars, recessions and bear markets typically start at the beginning of the term of a new President. There has not been a stock market loss in a pre-election year since 1939.

During most of the 20th century the US stock market has shown a cyclical four-year pattern of three up (bull) years followed by one down (bear) year. The exceptional periods have been 1929-1932, 1973-1974 and 2000-2002. Commodity markets (and countries with stable commodity-based economies) perform best during the stock market down years. Government bonds also typically perform better when stock markets are poor.

The so-called "January barometer" of the market may also be related to the American political system. A down-January had predicted a down-year and an up-January has predicted an up-year for every odd-numbered year since 1937, the only exception being 2001. Odd-numbered years are when new American Congresses convene in January and the new policies initiated -- good or ill -- may impact the economy for the entire year. The first five trading days in January are usually down in bear markets.

The so-called "January effect" is a January rally that particularly affects small "market cap" stocks. (Market capitalization is price per share of stock times the number of shares outstanding.) Since 1925 the return on small-cap stocks in the month of January has been four times the return on large-caps. And most of this effect is seen on the last trading day of December and the first four trading days in January. The explanation for the January effect is typically that many investors sell their worst stocks in December in order to take a tax-loss. Portfolio managers don't like to have people see stocks that have done poorly in their portfolios -- even if the prospects for those stocks look better for the coming year. Perhaps this effect favors small-caps because stocks that have dropped significantly in price have acquired a much smaller market cap (and more favorable P/E ratios).

Peculiarly there is also a "January effect" for so-called "junk bonds". Between 1985 and 2000 the Merrill Lynch High Yield Master Index averaged a 1.56% return each January, nearly double the 0.83% average for other months.

Ned Davis Research has demonstrated that from 1929-2000 a plot of the Russell 2000 divided by the Russell 1000 (small-cap divided by large-cap) shows a steady decline from June to December (except for late August/early September), a sharp spike upward in December/January, and a somewhat flat pattern until the beginning of June. In the 20-year period ended June 1997, large cap growth stocks outperformed large cap value stocks 72% of the time in the final quarter of the year. Value stocks (stocks with lower price-to-earnings ratios -- P/Es) performed best in the first quarter. Large-cap and small-cap value stocks beat large-cap and small-cap growth stocks an average of 68% and 58% respectively in the first quarter.

In the bull market of 1998 & 1999, the S&P 500 showed a gain of 26.67% & 19.53%, respectively. But without the best 5 trading-days, the returns would be 4.54% & 3.98%. Conversely, for the bear market in the first 10 months of the year 2000, avoiding the 5 worst trading days would reduce the S&P 500 loss from 18.03% to 0.92% [BARRON'S, 5-Nov-2000, page 20].

(return to contents)



(return to contents)