Dashboard de modelo predictivo de series temporales de activos de renta variable utilizando una red neuronal convolucional

Abstract

Abstract: The objective of this thesis is to implement a buy or sell indicator of a time series of a financial asset, through the use of a convolutional neural network. For this, the Python programming language and the Pytorch framework were used. The purpose of the project is to find a new financial indicator that can be incorporated into a speculation system in order to improve decision making and increase profitability. It was determined that the use of a CNN network helps reduce the impact produced by the noise in a time series, since its function allows identifying patterns in the data a more effective way. As a result, a new financial indicator was developed that quantifies; the error between the market price (which includes noise) and the price projected by CNN. This quantification is very useful in analysis and decision making in the market. financial, because it helps determine whether an asset is overbought or oversold, taking into account the price behavior in recent days.