The Greatest Stock Market Winners: An Investigation of Quantitative Investment Strategies based on the Greatest Stock Winners

Abstract

This thesis is investigating whether some of the most extreme stock market winnersshare any common quantitative characteristics that can be used to form a quantitative investment strategy. The motivation of conducting this research stems from the ambiguity in relation to the performance of some existing winner strategies formed by Marc Reinganum (1988) and William O’Neil (2009). Thus, the thesis seeks to provide updated perspectives on what drives the most extreme stock market winners and based on this, test whether these perspectives can be used to optimize the performance of quantitative winner strategies. The research is being conducted based on a sample of 1054 historical constituents of the S&P 400 Mid Cap index. The period was divided into three sub periods, 1994-2000, 2001-2009, 2010-2018, in which 63, 83 and 103 extreme stock performers were identified on the magnitude of their 12-months price acceleration. The analysis indicates that the identification and characterization of extreme stock market winners based on solely quantitative variables is highly sensitive to changes in the market conditions. As a result, the quantitative investment strategies are exposed to the different risk factors depending on what period the strategies are derived from. The strategies were out-of-sample tested during longer and shorter time horizons between 2001-2020 together with some alternative winner strategies based on existing stock winner literature. In most cases the winner strategies are delivering excess returns compared to the S&P 400 index. In the shorter periods, however, only a strategy derived from the existing winner literature provides statistically significant excess returns when adjusting for risk factor exposures. In the longer term from, 2001-2020, all tested winner strategies obtain yearly excess returns between 7%-11% compared to S&P 400, delivering statistically significant alpha after adjusting for risk factor. The results however are potentially exposed to look-a-head bias as result of the sample being constituted by historical constituents. When comparing the results with the performance of an equally weighted portfolio of all stocks in the sample, only 2 out of the 27 incidents of excess returns across strategies and time slots are statistic significantly higher than 0. It is concluded that the quantification of winner characteristics implies that performance of investment strategies is highly sensitive to changing market conditions. Moreover, the results indicate that most of the performance of the strategies can be explained by common risk factors, leaving investors with the decision of either exposing itself to these factors by active efforts of beating the market or more passively by investing in factor index-funds.