Analysis and optimization of a stock-picking model based on earnings momentum

Abstract

This project-thesis has been realized in the context of an internship at AlphaSearch. This company, based in Luxembourg, provides financial services by developing an application on Bloomberg that helps investors in their stock-picking process. AlphaSearch offers several factor-based models that seek to make profit of some market anomalies such as the price momentum, the earnings momentum and the value premium. The objective of the project-thesis was to analyze and to suggest an improvement to the earnings momentum model developed by AlphaSearch. Momentum investing is a stock investment strategy that aims to gain profit from trends existing on the market, with the belief that the trend will persist over time. Subsequently, earnings momentum is the tendency for stocks with growing earnings per share to outperform the market. Indeed, increasing earnings is usually a sign that the company is flourishing, which might be an indicator that stock price will increase as well. To carry out this mission, the first step was to look for better factors by building one-factor investment strategies. Practically, stocks are ranked from the best to the worst, according to the value of the factor. Then, a portfolio is built in which the top 10% stocks are bought and hold for 1 month. This strategy is repeated each month from 1999 to 2015. Based on the performance of these portfolios, the six best factors were selected to be integrated in the model. This first step is closed by the optimization of the weight given to each factor in the model. In the second phase, a backtest of the model was carried out in the U.S. universe to verify the efficiency of the model and to compare the performance of the new model to the initial one. This analysis was completed by fitting the earnings momentum model to Fama-French, Carhart and Asness’ pricing models, to explain the model’s exposure to risk factors.