Improving Pairs Trading Using Neural Network Techniques and Fundamental Ratios

Abstract

Pairs trading is a quantitative trading strategy consisting on identifying two stocks that historically move together and, using the assumption that their prices difference has mean- reverting properties, exploit the deviation from the mean by taking long – short position in the chosen pair to profit. Throughout the years, different approaches have been developed in order to exploit this strategy. However, there is little literature who looks whether the divergences in the prices are generated by poor company results, i.e. whether the deviation from the mean are product of bad (or good) fundamentals and are justified, or if they generate a new equilibrium point for the pair. In addition, since machine learning techniques are becoming more popular in finance, this work aims to analyze the performance of pairs trading strategy using neural network techniques applied to S&P 500 index components, selecting pairs of stocks from same industry and picking up the effects of the fundamental ratios in the pairs before taking a trade decision.

Keywords: Pairs Trading, Trading Strategy, Co-Integration, Mean-Reverting Process, Neural Network, Machine Learning, Fundamental Ratios

JEL Classification: G1