often quite predictable from one period to the next. Shouldn’t these pre-
dictable factors make stock prices move in nonrandom patterns?
In 1965, Professor Paul Samuelson of MIT showed that the ran-
domness in security prices did not contradict the laws of supply and de-
mand.^4 In fact, such randomness was a result of a free and efficient
market in which investors had already incorporated all the known fac-
tors influencing the price of the stock. This is the crux of the efficient mar-
ket hypothesis.
If the market is efficient, prices will change only when new, unan-
ticipated information is released to the market. Since unanticipated in-
formation is as likely to be good as it is to be bad, the resulting
movement in stock prices is random. Price charts will look like a random
walk since the probability that stocks go up or down is completely ran-
dom and cannot be predicted.^5
SIMULATIONS OF RANDOM STOCK PRICES
If stock prices are indeed random, their movements should not be dis-
tinguishable from counterfeits generated randomly by a computer. Fig-
ure 17-1 extends the experiment conceived by Professor Roberts 50 years
ago. Instead of generating only closing prices, I programmed the com-
puter to generate intraday prices, creating the popular high-low-close
bar graphs that are found in most newspapers and chart publications.
There are eight charts in Figure 17-1. A computer, using a random-
number generator, has simulated four of these charts. In these charts,
there is absolutely no way to predict the future from the past because fu-
ture movements are designed to be totally independent from the past.
The other four charts were chosen from actual data of the Dow Jones In-
dustrial Average over recent years. Before reading further, try to deter-
mine which are real historical prices and which are computer created.
Such a task is quite difficult. In fact, most of the top brokers at a
leading Wall Street firm found it impossible to tell the difference be-
tween real and counterfeit data. Only two-thirds of brokers correctly
identified Figure 17-1d, which depicts the period around the October 19,
1987, stock crash. With the remaining seven charts, the brokers showed
292 PART 4 Stock Fluctuations in the Short Run
(^4) Paul Samuelson, “Proof That Properly Anticipated Prices Fluctuate Randomly,” Industrial Manage-
ment Review, vol. 6 (1965), p. 49.
(^5) More generally, the sum of the product of each possible price change times the probability of its oc-
currence is zero. This is called a martingale, of which a random walk (50 percent probability up, 50
percent probability down) is a special case.