Incorporating Improved Directional Change and Regime Change Detection to Formulate Trading Strategies in Foreign Exchange Markets

Trading strategies for securing excess return and controlling investment risks in the FX (foreign exchange) market have attracted the attention of numerous researchers and investors. As a data-driven financial data sampling method, DC (Directional Change) can better capture price changes and perform well in the trading strategies of FX market. However, traditional trading strategies based on DC adopt the method of subjective selection of fixed DC threshold [[EQUATION]] , which depend on the trader's professional experience and have poor robustness. Moreover, the previous trading rules based on DC are limited to DC event itself, ignoring the grasp of the overall market dynamics. In order to overcome these defects, this paper proposes an intelligent trading algorithm (ITA) for FX market trading. The trading algorithm improves DC by adding attenuation coefficient [[EQUATION]] to the threshold of downward DC trend to make DC more sensitive to downward trend, and then optimizes super parameter combination [[EQUATION]] by Bayesian optimization. Furthermore, it introduces DC-related indicators as the input of HMM (hidden Markov model) to detect the regime change. In order to validate the effectiveness of ITA, we conduct a case study of three currency pairs: EUR/GBP, EUR/USD and EUR/JPY, which compares the proposed ITA with multiple fixed thresholds, dynamic threshold based on improved DC, regime change detection and no regime change detection. The results show that the trading strategy based on improved DC and regime change detection can obtain positive returns and a relatively low level of risk, which shows the effectiveness of the improved DC and regime change detection based on double threshold optimization in foreign exchange market transactions