In this paper I investigate whether historical stock prices in the Brazilian market have any power in predicting future returns. This would contradict the hypothesis that security prices follow a random walk. If future prices are random, technical analysis fails in identifying bullish or bearish trends. I estimate forecasting models of future returns based on historical returns, rank stocks by predicted returns and allocate them monthly in ten portfolios ordered by rank position. The portfolio with higher predicted returns significantly outperformed the market equilibrium, and the portfolio with lower predicted returns significantly underperformed the market equilibrium. This provides evidences that past returns carry some predictive power. One possible explanation has its roots in behavioral reasons.
Market efficiency; technical analysis; random walk; behavioral finance; trends in prices
