An Endogenous Feedback and Entropy Analysis in Machine Learning Model for Stock’s Return Forecast  

作　　者：Edson Vinicius Pontes Bastos Jorge Junio Moreira Antunes Lino Guimaraes Marujo Peter Fernandes Wanke Roberto Ivo da Rocha Lima Filho 

机构地区：[1]Instituto Alberto Luiz Coimbra de Pos-Graduação e Pesquisa de Engenharia-COPPE,Universidade Federal do Rio de Janeiro(UFRJ),Rio de Janeiro,21941-598,Brasil [2]Instituto de Pos-Graduacao em Administracao-Coppead,Universidade Federal do Rio de Janeiro(UFRJ),21941-918,Brasil

出　　处：《Intelligent Automation & Soft Computing》2023年第6期3175-3190,共16页智能自动化与软计算（英文）

摘　　要：Stock markets exhibit Brownian movement with random,non-linear,uncertain,evolutionary,non-parametric,nebulous,chaotic characteristics and dynamism with a high degree of complexity.Developing an algorithm to predict returns for decision-making is a challenging goal.In addition,the choice of vari-ables that will serve as input to the model represents a non-triviality,since it is possible to observe endogeneity problems between the predictor and the predicted variables.Thus,the goal is to analyze the endogenous origin of the stock return prediction model based on technical indicators.For this,we structure a feed-for-ward neural network.We evaluate the endogenous feedback between the pre-dicted returns and technical analysis indicators based on the generated residues.It is possible to predict the return.The high accuracy of the model indicates that,during the test period,there is a hit rate close to 76%.Regarding endogeneity,the term of interest and the return are the variables that influence the largest number of indicators.The results will help investors build investment strategies based on this expert system applied to forecasting.

关 键 词：FORECAST ENDOGENEITY neural networks differential evolution stochastic optimization 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]