An Effective Approach for Obtaining a Group Trading Strategy PortfolioUsing Grouping Genetic Algorithm(中文)

這篇論文利用10個不同的進場指標以及10個不同的出場指標組合成100種交易策略,接著透過篩選機制找到15支表現較好的策略。如何將這些策略分組並且分配資金可以找到報酬最高並且風險最小的投資組合就是GGA表現的地方。這份研究復刻了作者的方法並加以延伸,而這次使用到的標的是SPY和NASDAQ-100。
Grouping Genetic Algorithm (GGA) is a concept in the algorithm of computer science. In the paper An Effective Approach for Obtaining a Group Trading Strategy Portfolio Using Grouping Genetic Algorithm(2019), the author first combined GGA and trading and defines a group trading strategy portfolio (GTSP). It is designed to optimize a trading strategy portfolio, a set of strategies where the return and risk of the portfolio can be maximized and minimized, respectively.
This paper applies the concepts of the GGA algorithm to building portfolios. In the chromosome representation, the grouping, strategy, and weight parts are employed to encode a possible GTSP. In addition, the fitness value of a chromosome is calculated by the group balance (GB), weight balance (WB), portfolio return, and risk. The rest of the concepts are same to traditional genetic algorithm.
In this report, we try to use the GGA strategy on SPY and NASDAQ-100. We used the indicators provided by the authors to build 100 strategies for experimentation and extended them. The first is to change the method of chromosome selection and introduce the concept of Epsilon Greedy. Then adjust the way to calculate the fitness value, and observe how the group balance and weighted balance mentioned in the paper affect the final result.
Overall, in this experiment, we did not find the advantage of using GGA in the portfolio. If strategies are used as investment portfolios, we find that strategies will not be triggered most of the time, and the result is like putting capital directly in cash. Besides, the calculation method of GB and WB mentioned by the author will only make the result similar to put capital directly in each strategy equally weighted. The result is shown in Figure 1.
