A Study on the Application of Bandwidth Index in the Performance of High-Frequency Trading of Genetic Algorithm

Abstract

Continuous trading in Taiwan has received increased research interest in recent years. High-frequency trading is of particular interest. But, obtaining a high rate of return poses a lot of challenges. Many of these challenges can be overcome by algorithmic means. Genetic algorithms are an interesting class of algorithms for this kind of problem. Also, there is a growing body of literature that recognizes their importance. In this article we combined a genetic algorithm with a high-frequency trading perspective to consider bandwidth indicator application profitability. We used a simple genetic algorithm and develop an analytical model that examines the impact of different indicators on investment performance. We considered the following indicators: turnover rate, short-sale/margin ratio, bandwidth index and bid-ask ratio. Using these indicators, we select the highest-reward combination of stocks and the most appropriate trading frequency. Furthermore, we train and test the profit and loss performance of each stock group. Data for this study come from the Taiwan Stock Exchange Corporation (TWSE). We found the return rate of the bandwidth index is significantly better than other indicators. Five-minute intervals delivered the best trading results. Additionally, the findings show that the volatility of the bandwidth indicator has a high impact on the volatility of the return rate in high-frequency trading.