Handbook of Exchange Rates

12.4 Explaining The Success of Technical Analysis

Research from the 1970s through the early 1990s documented a long period, perhaps 15 years, in which fairly simple technical rules—MA and filter rules—apparently produced substantial excess returns on major exchange rates. This result was puzzling because it appeared to challenge the intuitively appealing weak-form efficient markets hypothesis, which stated that past prices should not produce positive risk-adjusted returns.

Several potentially complementary hypotheses have been put forward to explain the apparent success of technical analysis. First, there are at least three ways—data snooping, publication bias, and data mining—in which the apparent returns could be spurious, an artifact of the research process. Second, the apparent returns might be consistent with a reasonably efficient market if they are compensation for bearing risk. Third, either known agents—for example, central banks—or less understood factors—for example, cognitive biases—might introduce sufficient inefficiencies to create the observed returns.

12.4.1 Data Snooping, Publication BIAS, And Data Mining

Data snooping, publication bias, and data mining are three related but distinct problems that could tend to produce the false conclusion that TTRs are successful. In particular, these three issues stem from the fact that it is always possible to find ex post patterns, such as profitable trading strategies, that exist purely by chance in a specific data set and ...