Financial Post - Wednesday January 8, 2003
By George Bragues
It's the time of year when market analysts offer up their annual forecasts based on logic and evidence. They should leave it up to the psychics
'Tis the season for stock market predictions. Were we living in ancient times, seers would be trying to divine the future of the Dow Jones by interpreting a dream, entering into a frenzied communication with the gods, or examining the entrails of chickens. Living in more enlightened times, however, we now require that forecasters justify their claims more scientifically with logic and appeals to evidence.
Sad to say, this has done little to improve the quality of forecasts. To illustrate this, we need only dissect four of the more common rationales behind the plethora of 2003 market predictions.
The improbability of a four-year losing streak. In by far the most popular line of reasoning this year, bullish prognosticators observe that the market has declined three years in a row. Because the market only had one four-year run of losses during the 20th century, between 1929 and 1932, the chances of it happening again in 2003 are deduced to be slim. Writing in The Wall Street Journal, Jeff D. Opdyke points out that anyone who went long after any one of the 20th century's three-year losing streaks (1901-1903, 1929-1931, 1939-1941) would have enjoyed positive returns two out of the three proceeding years. This is about as logical as the sports gambling tout who tells us to bet a certain football team because it is three for four on grass after playing on Monday nights.
With only three instances of three-year losing streaks to consider, we really can't be sure whether the subsequent outcome indicates a genuine pattern or mere chance. A more significant error comes from ignoring the statistical fact that the stock market's performance in any given year is independent of what happened in earlier years.
Do the math on year-ending levels of the S&P Index between 1801 and 2002 (tabulated at
www.globalfindata.com), and you'll discover that the correlation between annual price changes and previous one-, three- and five-year returns is practically zero. Each new year, then, the stock market is comparable to a coin toss, with the only difference being that the coin is skewed slightly in favour of the upside. How much, we cannot be sure, though the proportion of positive years on the S&P over the last two centuries suggests the market should rise about 60% of the time. As such, the best forecast to make on Jan. 1 is always to be bullish, no matter what has recently transpired.
The Pre-Presidential Election Year Indicator. Floyd Norris of The New York Times is among many reminding us that, "it has been more than six decades since the third year of a presidential election cycle brought lower share prices." Trouble is, in all but one instance during the last 60 years, Britain's stock market also rose in a pre-presidential election year, despite that country not having a regular four-year cycle. So, too, this indicator only impresses if the historical evidence adduced is restricted to the post-war era. Between 1946 and 2002, a simple regression analysis -- a mathematical technique helpful in disclosing the predictive value of a stipulated variable -- reveals that the occasion of a pre-presidential year could indeed explain a 12% increase in the S&P index and do so with statistical significance. But when the same calculations were run for the longer 1801-2002 period, the incidence of a pre-presidential year ceased to be statistically significant.
P/E ratios. While the bulls monopolize the first two rationales, this one tends to be favoured by the bears, who insist that the market's P/E ratio, currently 30 on the S&P 500 index, continues to be well above the historical norm of approximately 15.
Academic studies give a decidedly mixed picture of the P/E ratio's ability to predict the market. The appropriate P/E ratio, after all, must take into account a myriad of factors, including earnings growth, market volatility, dividends, bond yields and inflation -- for these impinge on how much an investor should be willing to pay for each dollar of corporate earnings. In itself, as demonstrated by Princeton University's Robert Shiller in Irrational Exuberance, P/E only seems to be prophetic when an inflation-adjusted figure for the S&P is used in the numerator and a 10-year moving average of earnings is inputted in the denominator. Even then, it's only useful in forecasting the next 10 years, not the upcoming one.
Investor Sentiment. Long a favourite among contrarians, this indicator states that high levels of investor bullishness foreshadow declining stock prices, while pervasive bearishness augurs a rally. Accordingly, bulls appeal to the public's disillusionment with stocks, while bears espy a resurgence of investor excitement in the tech stock rally since October. In his book Beyond Greed and Fear, Santa Clara University professor Hersh Shefrin charts the percentage of bulls reported in the widely cited Investors Intelligence sentiment index against the subsequent year's market performance. No connection between the two comes to sight.
Given the strong public demand for predictions, it would be too much to ask financial journalists and brokerage analysts to stop issuing annual forecasts. But they could conceivably fulfill the demand by emphasizing particular stocks, where academic research has uncovered several predictive factors, such as average volume turnover, earnings momentum, valuation ratios, sales growth rates, as well as the level of accruals and capital spending in relation to assets. Otherwise, leave the annual predictions about the general market to the psychics.
I follow him since the street.com days A 1/3 of my account is piggy back with his trades with stops and that just add a little more profits to my account.