Research On Investment Portfolio Algorithm Driven By Risk Asset Price Informatio

Portfolio selection involves rational investment in a set of risky assets with the aim of maximizing investment returns and diversifying investment risks.Currently,researchers are drawing on emerging methods such as uncertainty theory and machine learning to propose multi-period investment portfolio studies and online investment portfolio research.However,these research methods often face challenges in obtaining analytical solutions for the models and lack targeted measurements for asset price trends,which can result in unsatisfactory investment returns.Therefore,this study proposes a risk measure,two price trend measures,and three portfolio selection algorithms based on risk asset price information,using uncertain theory and machine learning algorithms as research tools.Firstly,we propose the PRMP algorithm,a multi-period uncertain portfolio algorithm based on probabilistic risk measurement.We introduce a new probabilistic risk measure(PRM)that enables investors with different risk preferences to choose investment decisions with different risks without calculating the covariance of risky assets.We establish a bi-objective optimization model that maximizes expected returns and minimizes expected risks integrating PRM.We obtain an analytical solution by using equivalent transformation and linear programming methods,which we then convert into a portfolio.Empirical analysis shows that PRM is reasonable and effective,and the PRMP algorithm is practical and competitive.Secondly,we propose the APPTNPM algorithm,an online portfolio algorithm based on the adjusted peak price following winner strategy.Specifically,we first calculate the loss of predicting the current risk asset price based on the peak price of the previous time window,and use this predicted loss to adjust the highest price of the current time window to calculate the price trend,known as the APPT algorithm.By combining the APPT algorithm,we developed an optimization model that tracks investment decisions using adjusted peak prices,which we refer to as the APPTNPM algorithm.Empirical analysis shows that using the APPT to evaluate the price trend of risky assets is more advantageous than using peak prices.APPTNPM is competitive in terms of wealth accumulation and risk diversification capabilities,and is a robust and effective online investment portfolio algorithm.Thirdly,we propose the OLTR algorithm,an online portfolio selection algorithm based on the trend effect following winner strategy.We construct an index that measures the trend effect of risky assets and establish an optimization problem that tracks the trend effect and maximizes expected returns.We use the Lagrange multiplier method and soft projection optimization method to obtain a portfolio.Empirical analysis shows that OLTR outperforms typical benchmark strategies and follow winner strategies in wealth accumulation and risk diversification on five benchmark datasets.The proposed trend quantification method is effective.In summary,we analyze risky asset price information and establish targeted optimization models using appropriate optimization methods to obtain investment portfolios.Through scientific experiments and statistical verification,we obtain competitive experimental results,verifying the effectiveness of our research and promoting the study of investment portfolios.