Bollinger Bands Trading Strategy Optimization with Genetic Algorithm

Abstract

This work aims to apply a genetic algorithm to optimize a trading strategy based on the Bollinger Bands volatility technical indicator. The strategy defined here uses this indicator to trigger buy and sell signals, and different choices for the indicator parameters lead to different strategy final returns. A target optimization function is defined to maximize this final return by testing different combinations of the indicator parameters, and a genetic algorithm is modeled and applied to obtain optimal results. This approach is repeated with different time series in an intraday US stocks historical database, and the optimal strategy settings are then compared with the standard settings as a benchmark in subsequent periods, showing performance enhancement in around 80% of the cases.