A Comprehensive Price Prediction System Based On Inverse Multiquadric Radial Basis Function

Portfolio selection(PS)is a hot topic in modern financial field.Price prediction is the cornerstone of portfolio selection.At present,the existing price prediction strategies generally use a single prediction criterion or lack effective integration mechanism.In this paper,we propose a comprehensive price prediction(CPP)system based on inverse multiquadric(IMQ)radial basis function.Firstly,we design a novel RBF system based on IMQ radial basis function,which is centered on the portfolio vectors generated by multiple types of price prediction strategies,and uses IMQ radial basis function to measure the weight allocation of each strategy.In the selection of strategies,we consider both aggressive and moderate strategies,in order to achieve a better balance of benefits and risks in the volatile financial environment.Therefore,The novel RBF system can effectively integrate advantages of multiple strategies based on the current financial environment,making price prediction more stable and accurate.Secondly,based on the novel RBF system,a portfolio updating strategy based on kernel and trace operator is designed,and the solution is designed.In order to effectively evaluate the overall performance of the CPP system,we used four data sets from different financial markets and designed an experimental scheme with seven evaluation indicators in three aspects for a large number of experiments.The three aspects mentioned above are investment performance,risk measurement and practical application.Investment performance includes cumulative wealth(CW),average excess return(MER)andα factor,and risk measurement includes Sharpe ratio(SR)and information ratio(IR).In practical application,this paper tests the cumulative wealth and running time of PS system considering different transaction costs.The experimental results show that CPP system can achieve effective integration of advantages from different types of strategies,so that the system can better adapt to the complex market environment.In addition,the experimental results also confirm that CPP system is superior to other common PS systems in terms of investment performance,risk control and practical application.Finally,CPP system has the potential and ability to be applied to realistic large-scale financial environment due to its rapid calculation and short time consumption.