ترکیب ​​استراتژی های بازگشت به میانگین و تجارت تکانه در بازار ارز خارجی

The literature on equity markets documents the existence of mean reversion and momentum phenomena. Researchers in foreign exchange markets find that foreign exchange rates also display behaviors akin to momentum and mean reversion. This paper implements a trading strategy combining mean reversion and momentum in foreign exchange markets. The strategy was originally designed for equity markets, but it also generates abnormal returns when applied to uncovered interest parity deviations for five countries. I find that the pattern for the positions thus created in the foreign exchange markets is qualitatively similar to that found in the equity markets. Quantitatively, this strategy performs better in foreign exchange markets than in equity markets. Also, it outperforms traditional foreign exchange trading strategies, such as carry trades and moving average rules.

Foreign exchange market trading strategies have attracted much attention, especially since Fama (1984) introduced the “forward puzzle,” which argues that forward exchange rates are biased predictors of spot exchange rates. This paper sets forth a new strategy in the foreign exchange (FX) markets that combines mean reversion and momentum. Even though the strategy was originally designed for equity markets, I find that it produces higher Sharpe ratios than traditional FX strategies. The starting point of this paper, the “forward puzzle,” results from the rejection of the uncovered interest parity (UIP) theory. UIP states that the change in the exchange rate should incorporate any interest rate differentials between the two currencies. A large literature exists examining if and when UIP holds.1 This paper tries to find a pattern in the deviations from UIP and to explore the similarity between this pattern and that of stock returns. The literature reveals what looks like mean reversion and momentum in both markets. A long-run mean reverting pattern in currency values has been uncovered by Engel and Hamilton (1990); a short-term momentum effect generates profitability in FX market trading (Okunev and White, 2003).2Chiang and Jiang (1995) notice that foreign exchange returns show strong positive correlations in the short-run (momentum behavior) and negative correlations in the long-run (mean reverting behavior). This paper generates abnormal returns by employing a strategy that combines the long-run and short-run patterns of the deviations from UIP. The success of the combined momentum-mean reversion strategy brings about another interesting issue: the puzzling relationship between stock and FX markets. The similarities between the stock and FX markets are perplexing because macroeconomic fundamentals explain stock returns, but not exchange rates (Meese and Rogoff, 1983). Traditional theory dictates that two markets not relying on the same fundamental variables cannot behave similarly. Yet, I find the two markets comparable. This result is in line with early studies that depict similar empirical regularities in FX and stock markets (Mussa, 1979). One possible explanation springs from Engel and West (2005), who find that exchange rates explain macroeconomic fundamentals, not the reverse. This finding, corroborated with the finding that fundamentals explain stock returns, provides one possible channel for the relation between the two analyzed markets. Another fascinating explanation is that the risk factors affecting both stock and FX returns remain unknown but are somehow connected. An additional explanation is that similar behavioral biases operate in both markets, leading to similar inefficiencies. To explore the similarities between the stock and FX markets, I first consider the non-parametric approach that Jegadeesh and Titman (1993) exploit. These papers construct portfolio deciles based on previous months’ returns and choose a winner and a loser decile. By buying the winner and selling the loser a zero-investment portfolio is constructed and this portfolio is held for less than a year. The authors find that the return on their zero-investment portfolio is always positive. If this portfolio is held for more than a year, however, the return becomes zero or negative. The aforementioned non-parametric strategy is used in the FX market by identifying a winner and a loser currency based on previous deviations from UIP. I find that the winner continues to have high returns and the loser low returns for the subsequent 9–12 months, but, in the subsequent 4–5 years, the winner and loser portfolios switch positions. However, one cannot combine mean reversion and momentum strategies with this approach. Balvers and Wu (2006) use an alternative approach to generate trading profits in the stock market: a parametric strategy. They consider the effect of momentum and mean reversion jointly and conclude that the resulting strategy can lead to significant profits when applied to the stock markets of 18 developed countries. A contrarian strategy or a momentum strategy by itself leads to lower abnormal returns than the combination strategy. I find that the parameters obtained for the FX market are quantitatively similar to those for the stock market; hence we expect similar trading strategy returns for the zero-investment portfolios. The FX market returns have lower volatility than equity market returns. Consequently, I consider the Sharpe ratios, which allow me to compare risk-to-reward profiles of the same strategy, but in the two different markets. The Sharpe ratios obtained in the FX market are significantly larger than those obtained in the stock market. The paper is organized as follows. The second section describes the data and presents preliminary results showing that deviations from UIP exhibit momentum initially and subsequent mean reverting behavior. The third section describes the model and the fourth examines the model empirically and compares the results to the existing literature. The final section concludes.

This paper employs a trading strategy, previously applied only to the stock market that creates abnormal returns in the FX market. By running a simple parametric test, I find that UIP deviations follow mean reversion and momentum. In the FX market the half-life of mean reversion is very close to that obtained for the stock market, while the momentum effect is stronger than in the stock market. The combination strategy creates significant abnormal mean returns (slightly underperforming those of the stock market) and Sharpe ratios usually much higher than in the stock market. The results are also strong in comparison to strategies developed specifically for FX markets. Transaction costs do not alter the results significantly. I consider developed countries only. The portfolio I construct should be even more profitable if the currency choices are more numerous. This paper contributes to the literature not only by applying a new strategy in the FX market, but by applying one originally designed for the stock market. The FX literature considers mean reverting behavior toward PPP values and momentum trading strategies based on moving average rules. I bring a fresh perspective to understanding FX market dynamics by considering risky asset returns, instead of macroeconomic fundamentals. This allows for the creation of a strategy based on returns (computed in this case from deviations from UIP) and a direct comparison of the exploitability of the return patterns between the FX and the stock market. Up to now strategies employed in one market have not been successfully implemented in the other market, likely due to fundamental differences between the two markets. For instance, technical trading rules were found to be completely useless in the stock market since the publication of Fama and Blume (1966), but profitable in the FX market (Sweeney, 1986, Szakmary and Mathur, 1997 and LeBaron, 1999), suggesting major differences between the markets. The striking results here, on the other hand, raise the question of why the two markets behave so similarly. Many papers challenge the efficiency of the two markets. Neely (2002) examines the possibility that the FX market is inefficient due to Central Bank intervention. He reviews the literature regarding the abnormal returns obtained in the FX market through technical rules, but not existing in the stock market. Presumably, they are due to Central Bank interventions. Neely (2002) challenges that finding. This paper finds further evidence supporting Neely’s claim since the FX and stock market inefficiencies have similar patterns. One is left to wonder whether the two markets are indeed inefficient, or whether, in fact, there exists an unobserved risk factor that explains these returns. A possible theory is the overreaction hypothesis proposed from a behavioral perspective (e.g. De Bondt and Thaler, 1985) for the stock market: individuals tend to “overreact” to recent information, creating momentum. After some time, extreme movements in prices will be followed by a return to fundamentals, leading to movements in the opposite direction, creating mean reversion. In foreign exchange markets, Dornbusch (1976) shows that exchange rates tend to “overshoot” in their response to monetary policies, but then revert to a long-run equilibrium. The correspondence between overreaction in the stock market and overshooting in the foreign exchange market may be responsible for the similarity in results in these two markets when the combined momentum and mean reversion strategy is applied. Examining this correspondence presents a promising direction for future research.