Can Technical Analysis Predict Commodity Price Developments?

Casper Bek

Student thesis: Master thesis


The purpose of this thesis is to provide a stronger understanding of commodity markets by asking the question, is technical analysis a reliable tool for predicting future developments in prices of key commodities? Considering the recurring debate about the merits of the Efficient Market Hypothesis, this study is an exploratory study that seeks to test the hypothesis that technical trading systems have no predictive power in the commodity markets. According to the EMH there should be no predictive power of such a strategy since market efficiency ensures that all available information is reflected in today’s prices. Hence, past prices should provide no additional information about future prices if the market is efficient. Consequently, findings that indicate reliable predictive power of technical analysis also provide strong support for the argument that the commodity markets are not efficient in the weak form of market efficiency. In light of previous research on the EMH and technical trading, this study builds upon three separate methods to assess the research question. Where some of the previous research sometimes applied one or two methods on the same datasets, this study shows the strength of applying more methods on the same datasets. Each method adds a perspective on the data and the technical trading systems. Especially the application of both Monte Carlo simulations and comparisons of sub-period performance grant a strong support for accepting the null hypothesis. Using data for aluminum, copper, tin, and zinc prices from 1993 to 2013, the study finds that there is no predictive power of technical analysis. A review of the random walk properties of the data reveal that the prices in themselves differ significantly from the random walk hypothesis. However, the price changes exhibit some form of random behavior. Applying simple technical trading systems on the datasets show that some of the technical trading systems are able to make predictions about future market developments. However, applying those systems in EGARCH Monte Carlo simulations strongly suggests that the technical trading systems do not perform better than the market on a consistent basis. Identifying successful systems in one sub-period and then testing in other periods also underline that the systems are not able to perform consistently in- and out-of-sample. Consequently, the findings of the paper indicate that Burton G. Malkiel was right when he said, “Any successful technical scheme must ultimately be self-defeating.”

EducationsMSc in Finance and Strategic Management, (Graduate Programme) Final Thesis
Publication date2013
Number of pages133