Do Patterns Exist?

Some academics and investors believe that patterns do not exist. They believe either that price action is completely random or, at least, is indecipherable. We saw in Chapter 4, “The Technical Analysis Controversy,” that the concept of randomness is now being questioned, leaving open the possibility that order does exist in prices. However, even if order does exist in market prices, it is possible that it cannot be recognized with present mathematical models because it is so complex. The methods used in chaos theory, neural networks, and other esoteric mathematical models may prove useful sometime in the future, but not now. Thus, there is still the realistic question of whether patterns do exist in prices. Technical analysts swear they do, but in many cases, analysts are not mathematically sophisticated enough to demonstrate their validity.

As mentioned in Chapter 11, “History and Construction of Charts,” the unpublished article by Hasanhodzic et al. (2010) on a study of online video game players (http://arora.ccs.neu.edu) attempting to distinguish between financial market price statistics in moving chart form from random permutations of the same data with immediate feedback found that these people could “consistently distinguish between the two types of time series” (p. 1). This experiment seemed to give evidence that humans can learn to distinguish patterns and real data from random series whereas computers, so far, cannot.

If prices do have patterns, what causes them? This has been a debate for at least a century and has coalesced into a belief that patterns are the result of human behavior, which, conveniently, is indecipherable. It is, however, why technical analysts are very much interested in the new behavioral finance and neurological studies. They hope that the biases and tendencies in human behavior now being measured and gradually understood by behavioral finance students will eventually explain why price patterns seem to exist.

Behavioral Finance and Pattern Recognition

The first fact to acknowledge about chart patterns is that they have not been proven to exist or to be profitable. Although many investors and traders swear by certain formations, their evidence is largely anecdotal. Added to this is the tendency to see patterns in random data.

Humans have a tendency to want patterns in data and other information and to see them when they do not exist. Superstitions are derived from the erroneous and coincidental observations of patterns that do not exist but are created because of the desire to have a pattern. B. F. Skinner, a famous Harvard psychology professor, studied pigeon behavior in a number of stimulus-response situations to see if the pigeons would react to various stimuli and thus “learn” responses. The reward for the correct response was usually food. In one experiment, he decided to give pigeons just food without a stimulus to see what they would do. Invariably, in trying to make sense out of stimulus-response, the pigeons responded in different manners by creating their own stimulus, some bobbing, some developing strange head motions, thus creating their own superstition, when a real stimulus did not exist (Skinner, 1947).

Humans are similar in their desire to have some kind of stimulus, even if it is a black cat crossing the road, and develop supposedly predictive relationships when none actually exists. This is a special danger in price analysis because the desire to see a pattern can occur when no pattern actually exists.

Humans are also poor statisticians and tend to put more weight on recent history than what is statistically warranted. An experiment by Kahneman and Tversky (1982) showed that in flipping coins, which have a statistical probability of landing on their heads 50% of the time regardless of what side they landed on earlier, observers began to expect more heads in the future when the sequence of heads turned up more frequently, and they were surprised when that did not occur. Their subconscious brains expected more heads because that was the most recent history of flips, even though the odds had not changed. In the technical analysis of patterns, the analyst must guard against superstition or what is often called “market lore.” Frequently these statements contradict other statements or are just plain wrong. An example is that “a descending triangle always breaks downward.” You will see how this is not borne out by fact when we discuss triangles. Pattern and trend analysis must be based on evidence alone.

Humans also tend to see the future as the past and look backward rather than forward. This bias is likely the reason that trends in prices exist in the first place and why prices rise or fall until they reach some exhaustion limit rather than adjusting immediately as the EMH would suggest. For this reason, humans have difficulty in recognizing when past signs or patterns are no longer valid. Studies have shown that the human brain releases dopamine (a pleasure sensation chemical and, thus, a reward) when a human takes action that has worked before. Thus, the pleasurable action is desirable and overcomes any cognitive reasoning that might suspect the action is wrong. This problem is especially prevalent and potentially very dangerous in the financial markets where change is constant.

“In a world without change, the best way to find cheese is to return to the location where it was found previously. In a world with change, however, the best way to find cheese is to look somewhere new” Burnham (2005, p. 284), paraphrasing from Johnson (1998). In other words, the chart pattern and trend reader should look for failure rather than believe in the constancy of previous patterns. Schwager (1996) suggests that profitability from failed patterns is often greater than from correct patterns.

Source: Kirkpatrick II Charles D., Dahlquist Julie R. (2015), Technical Analysis: The Complete Resource for Financial Market Technicians, FT Press; 3rd edition.