Momentum Trading Results
We execute the following procedure to calculate momentum returns. The ranking period is defined as the period during which returns are calculated in order to rank the shares according to their relative return levels. The
holding period is defined as the period after the ranking period, during which the returns from holding the shares are realized. Every month, for a specific ranking period, the returns of all shares are calculated and ranked. The top five winners are bought and the bottom five losers are sold short, and we thus hold these 10 shares in position. We initially choose the minimum number of shares per long or short portfolio to equal five, as other momentum trading studies also often use deciles. Furthermore, we below analyze the effects of changing this number and show that a reduction increases the momentum returns generated but at the same time lowers the statistical reliability of the results.
We vary both the ranking and the holding period between 1 and 12 months. We utilize the investment rule that equally weighted momentum strategies of varying vintages are simultaneously in effect at all times, following Jegadeesh and Titman (1993, 2001) and Griffin, Ji, and Martin (2003). For comparison purposes, we apply the most commonly used return calculation method. We acknowledge that the implicit monthly rebalancing is potentially oversimplifying and may, in practice, increase transaction costs, as pointed out by Liu and Strong (2008). Moreover, we follow the common practice of skipping one month between ranking and holding periods in order to minimize microstructure price distortions. As a robustness test, we also calculated momentum returns without skipping one month. The results obtained contain few surprises and are broadly in line with, for example, the findings of Griffin, Ji, and Martin (2003), and are available from the authors upon request. For each of the five data samples, the annualized momentum returns are calculated. Table 3.1 reports the returns for the highest versus lowest portfolio across several ranking and holding period combinations. Corresponding t values are given in parentheses.
Table 3.1 shows for all five geographical areas that positive momentum returns are generated for the most commonly used ranking and holding period combinations of 6, 9, and 12 months. Apparently, a sample size of 30 to 60 shares suffices to convincingly demonstrate the pervasive presence of momentum effects. For three of the five samples (Europe, Nordic, and the United States), the returns are statistically not significant. For Australia and Canada, the returns are markedly higher and significant, as confirmed by the t statistics. For these samples, the annualized returns range between 20 percent and 30 percent for various ranking and holding periods.
chapter 3 Momentum Trading for the Private Investor Table 3.1 Momentum Returns
Average user rating: 5 stars out of 1 votes
Post a comment