《doesmomentummatter》由会员分享,可在线阅读,更多相关《doesmomentummatter(4页珍藏版)》请在金锄头文库上搜索。
1、Does Momentum Matter? - Using Daily Returns to Create TradingStrategies for Internet StocksGruezi Mitenand Capital, Ltd.Matthias HoffmannBernardo MartinezJuan Pablo de MiguelSergio PenchasWill WalkerDuke UniversityThe Fuqua School of BusinessBA 453 International InvestmentsProfessor Campbell R. Harv
2、eyOVERVIEWThe purpose of the following analysis is to test the effectiveness of various trading strategies for a broadly based sample of internet stocks. We constructed a series of portfolios, each adjusted daily for movements in the prices of its component stocks. By combining long and short positi
3、ons in these various portfolios, we were able to determine whether there exist in the market opportunities for profits that can be recognized by monitoring daily stock returns. “Effectiveness,” for the purposes of this discussion, is determined by measuring the volatility and return of each particul
4、ar portfolio versus those for a buy-and-hold portfolio of a synthetic, equally-weighted internet stock index. Two definitions should here be introduced. For a “momentum” strategy, after each days trading, two portfolios are created one including the days highest-returning performers, the other conta
5、ining the days lowest-returning performers. The following day, the “top” portfolio is purchased, the “bottom” portfolio is sold short. At the end of that day, both accounts are closed out, and the evaluation process begins again for the next days transactions. A “contrarian” strategy is executed the
6、 same way, except the days top performers are sold, and the bottom performers are purchased. METHODWe collected daily stock prices for internet companies over the period from September 17, 1997 to December 31, 1999. We obtained these historical quotes from Whartons Center for Research in Security Pr
7、ices and from Yahoo. We adjusted our daily returns to take transaction costs into account by incorporating a bid-ask spread of 0.30% and a nominal transaction fee of $6. We chose to begin our analysis in September, 1997 because, at that time, forty-eight internet stocks were trading in the market, p
8、roviding us with enough stocks for our momentum strategy analysis to be feasible and relevant. For subsequent internet IPOs, we included the newly-traded stocks chronologically in our possible selection basket. Likewise, as internet stocks were taken off the market, we adjusted our model to ignore t
9、hose stocks one day prior to their last trading day. By December, 1999, 143 internet stocks were trading. To see the growth of the internet market, please refer to Exhibit One, “Number of Stocks in the Sample.”We selected the ten top and the ten bottom stock performers for each day, effectively crea
10、ting two portfolios. We “invested” $5 million per stock in each portfolio, yielding a $50 million value per portfolio. Given the top and bottom portfolios for each day, we essentially were faced with six possible strategies: (1) buy the top, (2) sell the bottom, (3) momentum (buy the top and sell th
11、e bottom), (4) sell the top, (5) buy the bottom, (6) contrarian (sell the top and buy the bottom).For comparison, we constructed a benchmark index, an equally-weighted portfolio of available internet stocks for each day. We then calculated a cumulative benchmark return, measuring the return over the
12、 relevant time period from a buy-and-hold strategy for the index. This return and its corresponding volatility could be compared with the returns and volatilities for our six possible strategies.RESULTS As the summary table below indicates, with the exception of the “Buy Bottom” strategy, all our st
13、rategies underperformed with respect to the benchmark, in terms of both return and volatility. For a time-series plot of cumulative returns, please see Exhibit Two, “Cumulative Return.”PortfolioBuy TopSell Bott.Moment.Buy/HoldContrar.Sell TopBuy Bott.Betas1.19 (0.95)0.24 1.00 (0.24)(1.19)0.96 Averag
14、e0.17%-1.02%-0.85%0.37%-0.14%-0.66%0.53%STD4.18%3.56%4.47%2.53%4.47%4.17%3.56%Min-16.06%-16.89%-19.10%-12.79%-28.24%-33.59%-13.76%Max33.23%13.22%27.40%11.28%18.02%15.76%16.29%The Buy Bottom strategy generated a higher return than the benchmark (0.53% daily versus 0.37% daily), yet it also had a corr
15、espondingly higher standard deviation (3.56% versus 2.53%).DISCUSSIONOur results suggest that a Buy Bottom strategy is preferable with respect to internet stocks. Can this outcome be explained with respect to the nature of the internet market? Since 1997, internet stocks have represented a highly volatile, highly rewarding class of equities. Analysts have had difficulty valuing these companies because many will not see any earnings for a number of years. Our results suggest that, given the acu