Outperforming the S&P 500: A Computational Approach to Dynamic Portfolio Optimization

Abstract

COVID-19 dramatically increased the attractiveness of high-growing tech stocks and low-priced value stocks, making them the obvious choices for investors in 2020. As a result, a large population of retail investors joined the market and implemented this strategy, hoping to reap these stocks' large returns. In 2021, however, as the vaccine continues to roll out under the Biden administration and the market reverts back to pre-COVID conditions, this approach will no longer be as effective, leaving these investors without an obvious investment strategy. Therefore, this paper explores the historical data of the stock market, specifically the S&P 500 between 1989 and 2018, to find a strong, simple investment strategy which is relatively safe and outperforms the S&P 500's average return. It expands upon Goetzman and Ibbotson's (1995) "repeat-winner" hypothesis by examining the year-end returns for all of the constituent companies in the S&P 500 and allocating wealth into the top 10 performing ones based on annual percent returns to create the stock portfolio. This algorithm is replicated every year, selling current companies in the portfolio if they fall out of the top 10 and including new companies that have high return performance. Diversification, optimization, and risk are important factors to consider as well when creating an investment portfolio so the second half of the research includes corporate bonds and 10-Year US Treasuries into the portfolio and uses a mean-variance and mean-deviation Markowitz model to create optimal efficient frontier curves. Using these results, the paper concludes by comparing the risk of the S&P 500 at specific return levels with that of our investment strategy to determine if it is an effective strategy for this new population of investors.