This study examines the dual long memory properties in Malaysian stock market during bull and bear periods for the period 1993:10 to 2009:12. Both semiparametric and ARFIMA-FIGARCH models are applied for the diagnosis of long memory. The study finds no evidence of long memory for stock returns. On the contrary, the long memory in stock volatility for all the bear periods and three of the bull periods is supported. Besides, the long-range dependence is more persistent in the early years of the sample, in particular, prior to the imposition of capital controls by the Malaysian government in September 1998. The presence of long memory in volatility provides evidence against the efficient market hypothesis and thus offers arbitrage opportunities to reap excess profits in stock market.

After a series of bad news which led to the plunging of the US stock price index from 13,930 points (October 2007) to as low as 9,325 points (October 2008), many stock markets worldwide, then, also experienced a downturn shift. As a consequence, individual and institutional investors attempted to investigate the behavior of stock return series, particularly the long memory property.

Long memory property indicates not only the violation of the market efficiency, but its implications on technical trading rules. The weak form of the Efficient Market Hypothesis (EMH) asserts that the stock market is efficient based on past prices information (Fama, 1970) and thus, stock price should follow a ‘random walk’ or a process with no memory. Long memory property, on the contrary, indicates that the arrival of new market information cannot be fully arbitraged away (Mandelbrot, 1971; Granger and Joyeux, 1980; and Hosking, 1981). In other words, considering that what happens today may affect the future, if the information is fully utilized, there is a possibility to outperform the market and make consistent speculative profits. Therefore, the validity of EMH is questionable when current data is correlated with all past data in varying degrees.

Technical trading rule is built on the conception that the price movements follow a trend and are not random; thus, the movements are often predictable. Moving average, in particular, is one of the popular technical trading tools used by institutional investors and practitioners when proposing profitable strategies. With the presence of long memory component in stock prices, a higher-order moving average trading rule can be recommended to gain abnormal profits in the stock market.

Applied Economics Journal, Dual Long Memory, Stock Market Prices, New Evidence Based, Bull and Bear Markets, ARFIMA-FIGARCH, Efficient Market Hypothesis (EMH).