Data analysis in deep learning classification models, a financial application for bitcoin

Abstract

Uses for machine learning methods have dramatically increased over the last decade. With a diverse array of industries making use of it, it is no surprise that the financial industry has been one of its first adopters and pioneer in its development. However, precise measurements must be considered when dealing with financial data extracted from the market. This work project is an execution of Professor Marcos López de Prado (Cornell University)data analysis techniques for financial machine learning algorithms. The prepared data was then used as an input in a deep neural network for multi class classification, with the objective of making price direction predictions. Bitcoin was the selected financial instrument for this study, given its high volatility and its virtually global accessibility.