Forecasts and identification of buying and selling moments using time series models and neural networks in financial market

• Grant number: 21/04729-1
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: June 01, 2021
• End date: January 31, 2022
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Statistics
• Principal Investigator: Pedro Alberto Morettin
• Grantee: Augusto Kira Pedroso de Lima
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 18/04654-9 - Time series, wavelets and high dimensional data, AP.TEM

Abstract

Stock prices usually behave as random walks and ARMA-APARCH models are usually fitted to the log returns due to their heterocesdasticity. Forecasts for different horizons with confidence intervals will be based on APARCH models and exponential smoothing techniques and, without intervals, using neural networks, as the LSTM. Such forecasts will be used to identify trends of growth and fall in the price of stocks to identify moments of buy and sale. Models will be chosen based on their predictive ability, ie, all forecasts will be evaluated based on the mean quadratic error of forecast and better identification of the buying and selling moments to obtain greater returns.