Tests of the Random Walk Hypothesis Against a Price-Trend Hypothesis

Abstract

Forecasts of financial prices, calculated from the present and past values, are never substantially more accurate than the prediction that future prices will equal the most recently observed price. This conclusion has been summarized by two hypotheses. First, the random walk hypothesis, states that daily returns are uncorrelated. Few people believe that this is exactly correct; indeed, the hypothesis has been refuted both for small stock markets [13] and for American commodity futures markets [3]. Second, the weak-form efficient market hypothesis, as defined by Jensen [15], states that investors cannot make profits from any correlation between returns, after deducting all the costs of trading and adjusting for risk. Strict efficiency, is a special case that occurs when prices fully reflect all information available in the past prices [7], [8].