Research On Trading Strategy Based On Improved Directional Change And Regime Change Detection

Transactions do not occur at equal intervals.Therefore,financial UHF data observed are also not equally spaced.Most previous researches sampled data based on the equal time interval.However,due to the irregularity of the time interval of transactions,processing data with equal time intervals would lose important price changes,resulting in the lack of data.DC(Directional Change)and regime change detection are effective means to analyze price time series at present.DC is a data-driven financial data sampling method.DC samples when the price changes reach a certain threshold,and divide the market into upward DC trends and downward DC trends.In DC,threshold θ is the key to generate DC sequence,and it can determine the number and density of the DC events.Compared with the equal time interval approach,DC can better capture price changes.However,the traditional method of subjective selection of fixed DC threshold depends on the trader’s professional experience and has poor robustness.In this paper,DC is improved by adding attenuation coefficient α to the threshold value of downward DC trend to make DC more sensitive to downward trend.In addition,Bayesian optimization is used to find the optimal super parameter combination(θ,α)by maximizing the target return function,and it is applied to the subsequent trading.In addition,for the domain of regime change detection,Hidden Markov model has been widely used in many financial types of research.It has been proved that it can effectively detect the change of data structure and has been applied in the area of regime change detection.However,in the past,the research on the regime change was based on the equal interval time series data,which would destroy the internal structure of the original data and lose data information.In this paper,DC-related indicators are used to detect the regime change.By inputting the indicator series into the hidden Markov model,the hidden state of the market is found,and the market is divided into normal regime and abnormal regime with high volatility.In the abnormal state of the market,we stop trading to avoid risk.Based on the above research,this paper proposes ITA combined with improved DC and regime change detection and applies it to FX market trading.The results on tick data of nearly 200 million foreign exchange currency pairs 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.