摘要：The purpose of this paper aims to get empirical evidence about the performance of Technical analysis considering profitability when coupled on a stand-alone system to trigger orders of buy/selling shares via home broker at Bovespa, also present a modified methodology for the Lorenzoni et al. (2007) test to check the presence of informational content in technical analysis. This autonomous system consists of a modified routine of rational choice constant in Arrow (1959) which seeks maximum profit trying to simulate human behavior as a strategy for triggering orders. It was tested 352 different combinations of possible orders when attached to the rational behavior for each series tested. The data used were companies listed in FGV-100 index (theoretical portfolio of the year 2009) and Ibovespa (theoretical portfolio of Sept/Dez 2009) with a total of 129 companies from 3/01/2000 to 09/30/2009. The main results show that the best results were obtained by moving average followed by Bollinger bands and, finally the histogram, which has not proved by self efficient. Orders were counted in quantities. It was found rational choice behaviors of buy/sell with superior performance to its peers when compared in terms of profitability with the returns obtained by Ibovespa. The statistics of White (2000) and the modified test of Lorenzoni et al. (2007), proposed by the author, had suggested the presence of informational content for the majority of the series tested in the sample when the graphs indicated long position for both the original data and for the recreated data from the AR(1)-GARCH(1,1) model.
其他摘要：The purpose of this paper aims to get empirical evidence about the performance of Technical analysis considering profitability when coupled on a stand-alone system to trigger orders of buy/selling shares via home broker at Bovespa, also present a modified methodology for the Lorenzoni et al. (2007) test to check the presence of informational content in technical analysis. This autonomous system consists of a modified routine of rational choice constant in Arrow (1959) which seeks maximum profit trying to simulate human behavior as a strategy for triggering orders. It was tested 352 different combinations of possible orders when attached to the rational behavior for each series tested. The data used were companies listed in FGV-100 index (theoretical portfolio of the year 2009) and Ibovespa (theoretical portfolio of Sept/Dez 2009) with a total of 129 companies from 3/01/2000 to 09/30/2009. The main results show that the best results were obtained by moving average followed by Bollinger bands and, finally the histogram, which has not proved by self efficient. Orders were counted in quantities. It was found rational choice behaviors of buy/sell with superior performance to its peers when compared in terms of profitability with the returns obtained by Ibovespa. The statistics of White (2000) and the modified test of Lorenzoni et al. (2007), proposed by the author, had suggested the presence of informational content for the majority of the series tested in the sample when the graphs indicated long position for both the original data and for the recreated data from the AR(1)-GARCH(1,1) model.
