Arbitragem com Fundamentos Latentes em um Modelo Fatorial de Efeitos Mistos

We propose a single-factor mixed effects panel data model to create anarbitrage portfolio that identifies differences in firm-level latentfundamentals. Furthermore, we show that even though the characteristics thataffect returns are unknown variables, it is possible to identify thestrength of the combination of these latent fundamentals for each stock byfollowing a simple approach using historical data. As a result, a tradingstrategy that bought the stocks with the best fundamentals (strongfundamentals portfolio) and sold the stocks with the worst ones (weakfundamentals portfolio) realized significant risk-adjusted returns in theU.S. market for the period between July 1986 and June 2008. To ensurerobustness, we performed sub period and seasonal analyses and adjusted fortrading costs and we found further empirical evidence that using a simpleinvestment rule, that identified these latent fundamentals from thestructure of past returns, can lead to profit.