An effective technique for automatic portfolio stock selection, diversification, and optimization

Abstract In this paper, a novel stock-to-sector-to-benchmark ratio or anomaly that assists investors in automating stock selection, diversification, and optimization for portfolio management is introduced. The approach uses three market capitalization values, one for the individual stock, one for the sector, and one for the benchmark capitalization, to calculate the ratio. The results of this paper prove the efficacy of the proposed methodology. Out of the eleven constructed for the period under study, all the portfolios constructed beat the benchmark in terms of the highest weighted returns, lowest risk, and highest Sharpe ratio during the sample period (1979 to 2019). Fama‒French three-factor and five-factor models are used to assess whether the factor loadings influence performance. Although the Fama-French three-factor models showed a statistically significant alpha, the asset pricing model had an average adjusted R2 of 13%, while the adjusted R2 of the Fama-French five-factor model had an average of 60% (excluding a single stock-based portfolio). These portfolios exhibit a statistically significant negative SMB, which implies that the performance of the portfolios is affected by large capitalization stocks, which is in direct contrast to the popular belief that outperformance is influenced mainly by small-cap stocks.