Success and Failure of Technical Analysis in the Cocoa Futures Market

Abstract

A large set of 5350 trend following technical forecasting rules is applied to LIFFE and CSCE quoted cocoa futures prices, and to the Pound-Dollar exchange rate in the period 1983:1-1997:6. We find that 14% of these forecasting rules when applied to the LIFFE cocoa futures price earn a statistically significantly positive excess return, even when correcting for transaction costs. Moreover it seems that a large set of forecasting rules can distinguish periods with positive returns from periods with negative returns. On the other hand the same set of forecasting rules performs poor when applied to the CSCE cocoa futures price, with only 0.30% earning a statistically significantly positive excess return. Parametric bootstrap techniques reveal that the effectiveness of technical analysis in the LIFFE cocoa futures market can not be explained by characteristics as captured by several popular data generating processes like a random walk, autoregressive and GARCH model. However, the effectiveness of technical analysis can be explained by a structural break in the long-term price trend. The large difference in the effectiveness of technical analysis may be attributed to a combination of demand and supply of cocoa beans and an accidental influence of the Pound-Dollar exchange rate, strengthening price trends in the LIFFE cocoa futures price, but weakening price trends in the CSCE cocoa futures price. Our case study suggests a connection between the success or failure of technical analysis and the relative magnitudes of volatility and autocorrelation in returns, and trends in price of the underlying series.

Keywords: technical analysis, parametric bootstrap techniques, commodity futures, exchange rate

JEL Classification: G10