Análise de estratégias de trading utilizando técnicas de processamento digital de sinais

Abstract

Financial markets are essential in modern societies as they enable to trade assets quickly and at low transaction costs. One of the main players of these markets are known as traders, which are agents that assume risk and provides liquidity to traded assets. These agents seek profit by identifying of market inefficiencies. Traditionally, prospecting these phenomena come from graphical analysis. However, this approach does not allow an application of a systematic methodology for validation. Alternatively, the tools coming from the digital signal processing theory arise, that exploits statistical features of the financial time series, which provides the application of objective performance validation methodologies. One of the most sophisticated trend indicators presented in the literature, Kaufman’s adaptive moving average (KAMA), is based on adaptive filters theory. In this research, the performance of the KAMA indicator is compared to that of traditional trend indicators. The benchmark was made using financial industry standard metrics and representative assets of diverse market sectors. The obtained results indicate that the filter KAMA presents best results when applied in situations where the asset have smooth volatility and without “outbreaks”. In this case, one of the strategies which used KAMA filter achieved a Sharpe ratio of 1.48 for the asset S&P 500. On the other hand, using the same strategies when applying traditional indicators reached a Sharpe index between 0.97 and 1.25. Moreover, the results suggest that volatility can be interpreted as a measure for information processing of the market – what can justify “volatility outbreaks”.