Objective: to demonstrate the DC trading strategies can produce at least 12% yearly returns and formulate the best DC trading strategy by comparing the results between different optimization algorithms
Most of the financial forecasting methods use a traditional physical time scale, which is taken at fixed intervals (interval-based scale). In this project, an intrinsic time (event-based scale) using Directional Changes (DC) is applied to focus only on important points that a crucial event happened, filtering out other irrelevant price details and noise. In addition, this event-based scale is suitable to use in high frequency data. Using a traditional interval-based scale to study price fluctuations makes the flow of physical time discontinuous, so important price movements happening in several minutes cannot be captured. However, the intrinsic time using DC records all the key events in the market. In this project, we will take advantage of directional change paradigm to formulate different DC trading strategies.
In this mid-term report, single-threshold DC trading strategy and multi-threshold DC trading strategy have been currently developed and used to generate preliminary results. In the future, the DC trading strategies will be combined with different heuristic algorithms which are genetic algorithm (GA), virus spread optimization (VSO) and particle swarm optimization (PSO) to optimize the parameters. The goal of this project is to demonstrate DC trading strategy can make profitable returns and find out the best DC trading strategy by comparing different heuristic algorithms.
Please refer to the Trading Optimization using Directional Changes and AI Mid-Term Report.pdf for details