Alternative approach to characteristics-based investments

  • José Miguel Loreto Baptista Sismeiro (Student)

Student thesis: Master's Thesis

Abstract

Previous studies have shown significant return predictive ability of specific firm related variables, however, most of them only test one or a few variables at the same time. With the objective of exploring this shortcoming, this thesis analyzes the out-of-sample predictive ability of 12 firm characteristics to forecast returns. Forecasts are computed through cross-sectional Fama-Macbeth-style regressions. Moreover, we adopt 5 different combinations of characteristics in order to test the combined predictive ability of characteristics in each set. Additionally, we consider 4 estimation periods for the slopes derived from cross-sectional regressions. The main objective of this thesis is the development of a profitable firmcharacteristics related strategy for small investors. Therefore, we take into consideration 3 important and real constraints faced by small investors: i) a limited value available to implement the strategy, ii) no possibility of short-selling and iii) the payment of transaction costs. We find high predictive ability in all combinations, and among all lengths of estimation periods under analysis, which leads the investment strategy to outperform the market throughout the sample, regardless of the combinations tested. However, the combination of all 12 characteristics and 1-year rolling slopes stands out, leading the investment strategy to yield an average monthly return of 2% net of transaction cost and an annualized Sharpe Ratio of 0.96 contrasting with 0.39 from the S&P500. The strategy transforms $20,000 invested in 1982 into $28,843,056 at the end of 2015.
Date of Award21 Oct 2016
Original languagePortuguese
Awarding Institution
  • Universidade Católica Portuguesa
SupervisorBarbara A. Bukhvalova (Supervisor)

Designation

  • Mestrado em Finanças

Cite this

'