n the continual battle to make money in the stock market and futures markets, I many traders have found themselves in one of two opposing groups-the "technical analysts" or the "random walk theorists." Technical analysis rests on the belief that recent price, volume, and open interest movements signal future price changes; it is based on "the fact, which no one of experience doubts, that prices move in trends and trends tend to continue until something happens to change the supply-demand balance" (Edwards and Magee, 1966). A random walk theorist, on the other hand, believes that price fluctuations are totally random, and "cannot be predicted. . .the sad fact is that you are likely to be just as well off picking stocks by letting your fingers take a random walk down the Wall Street stock pages as by following [technical analysis]" (Malkiel, 1973).
The contradictory nature of these quotes accurately reflects the adversarial posture that technical analysts and random walk theorists have taken. It is not the purpose of this article to heal this rift. The techniques proposed here would most probably be used by a technical analyst, but could just as well be used by someone to show that technical analysis does not work.
A cornerstone of technical analysis is the use of timing devices, which give signals as to whether a commodity's price is going to increase or decrease. These signals are thus used to determine whether a short position or long position should be taken. They can also be used to determine if a current position should be liquidated, or no action at all should be taken, but we will only consider the simpler case of two possible signals.
There are many different types of timing devices, but all share the property that they use previous information about price and volume of a commodity to predict future price changes. Among the popular techniques are:
(a) filter system-a commodity that has reached an extreme point and then moved in the opposite direction 5% is considered to be in a trend