Keim suggests that what's been missing from past studies is an examination of price impacts triggered by real-life momentum trading; rather, profitability claims have been based on simulated momentum trading. "The simulated strategies are very mechanical and don't account for the impact the simulated trades would actually have on prices," Keim says. "In effect, the simulated strategies are buying and selling at prices that are not possible for actual traders. That's why it's so critical to look at the experience of actual trades."

The Wharton professor's study uses the actual costs of implementing a momentum strategy by examining more than 1.6 million institutional trades worth $1.1 trillion conducted in the Unites States and 36 other developed and emerging equity markets. These trades were placed during two 12-month periods between 1996 and 2000. Keim also compared the price tags of momentum trading with the costs generated by institutional investors pursuing either a value or diversified stock strategy.

The Plexus Group, a consulting firm that advises institutional investment managers and their traders on how to contain and control their trading costs, provided the trade data without revealing the identities of its clients.

According to Keim's study, momentum managers buying stocks in rising markets experienced a one-way average price impact of 1.89%. That means the transaction cost 1.89% more than the price of a typical retail trade. In comparison, the price impact was 0.91% for diversified traders and 0.85% for value managers. For momentum managers who were unloading stocks in falling markets, the price impact was even higher-2.24%. The cost of trading grew pricier when Keim segregated the trades by degree of difficulty, such as market capitalization and trade size. In the upper ranges of equity trade size, one-way price impacts for momentum adherents reached 3% to 3.5%.

Keim derived the price-impact figures through a mathematical formula that measured the financial impact of institutional trades. The formula is based on the premise that institutional trades are large and, consequently, the trades move security prices. For instance, an institutional trader wanting to buy 50,000 shares of a stock will have to pay a price concession to a market maker to complete the trade. The trader won't be able to buy that many shares at the prevailing ask price. The price impact equation measures this "excess" stock price movement.

"The price impacts here represent a clearer picture of the costs of implementing momentum strategies than previously reported in the literature and set a very high hurdle rate for the profits implied by the simulated strategies," Keim states in his study. "The reality inevitably falls short of the illusion."

While Keim believes that the returns of simulated momentum-based trades can't cover the costs of real-life momentum trading, he says one significant question remains unanswered: Can momentum traders in practice generate returns that exceed actual implementation costs? He acknowledges that the continued proliferation of momentum managers suggests that making money on this strategy is certainly possible.

Clifford S. Asness, a managing principal at AQR Capital Management LLC in New York, which uses momentum trading techniques, says the flaw in the recent research is in assuming that professional momentum traders only embrace that one particular investing style. "We at AQR, and many people who trade momentum, do not trade it alone in a vacuum. We trade it in combination with value strategies."

Asness cites as an example a long-short manager who divides his portfolio into a momentum-based long-short portfolio and a value-based long-short portfolio. Rather than trade them separately, he trades the net of the two positions. He gains two tremendous benefits from this approach. First, before transaction costs are calculated, both momentum and value approaches, enjoy positive historical Sharpe ratios and, even better, are negatively correlated with each other. Thus the two together have a much higher gross Sharpe ratio than either alone.

Second, transaction costs are greatly reduced by trading them together and are far lower than they would be in a pure momentum portfolio. Many securities that are good on momentum (or vice versa), but bad on value are simply never traded.