Volume Based Portfolio Strategies
Volume Based Portfolio Strategies
5 Results: Time-Stability of Portfolio Returns (p. 189-190)
In the last chapter, we identified a systematically positive relationship between abnormal volume and expected returns in the cross-section of Swiss stocks. In a next step, we investigate the robustness of portfolio returns across time and different market regimes, which is our research question . This set of analyses commences the testing of the practicability of abnormal volume based portfolio strategies, which continues in chapter 6 (investigation of the economic significance of portfolio returns).
The focus of this chapter is abnormal volume, because it is the only measure investigated shown to systematically and significantly relate to the cross-section of Swiss stock returns. However, we also briefly analyze the time-stability of the other volume measures, namely volume level, volume growth, and variability in volume. The reason is that there exists the possibility that these other volume-return relations are only systematically significant in specific states of the market, which would be an interesting finding by itself. The methodology applied is described in detail above, 3.3.2. Nevertheless, we repeat the most important aspects as deemed necessary.
5.1 Abnormal Volume
The analysis of the time-stability of the abnormal volume effect is divided into three parts, the dependence of results on few months with extreme returns, the dependence of results on returns of specific calendar months, and the analysis of portfolio returns across different market regimes. Before reporting test results, there are two methodological remarks regarding reference period and portfolios investigated: Reference period investigated: the analysis so far shows that the positive relationship between abnormal volume and expected returns is systematically significant across different reference periods investigated.
However, the portfolio returns are increasing in the reference period and consistently the highest when measuring abnormal volume compared to a long-term reference period of J = 12 months. We therefore restrict the analysis going forward to the J = 12 months reference period. Portfolios investigated: first, we again analyze quintile and decile portfolios based on one-way sorts on abnormal share turnover. Regarding two-way sorts, recall two important findings in 4.2.2. First, we showed that the abnormal volume effect does not proxy for previously discovered return determinants such as company size, book-to-market ratio, momentum, or industry affiliation.
And second, we discovered particularly profitable portfolio strategies within specific control variable quantiles, which is interesting regarding the practical implementation of volume based portfolio strategies. We thus also discuss the time-stability of abnormal volume deciles within the smallest one-third of firms per month (see Table 4.42, Panel C), within the one-third of firms with the lowest book-to-market ratio (see Table 4.43, Panel C), and within those one-third of stocks with the lowest past one-year returns (see Table 4.44, Panel C). Due to the diversification issue within industry groups, which by construction does not exist for other control variables, we abstain from individually testing abnormal volume decile portfolios within specific industry groups regarding practicability.