INVESTOR SENTIMENT AND THE STATE-DEPENDENT STRUCTURE OF FOREIGN CURRENCY MOMENTUM

    DOI: https://doie.org/10.65985/APER.2026202783

    Authors:

    Sheng-Hung Chen, Chun-Hsiung Hung, Pei-Shan Wu,


    Keywords:

    Investor sentiment, Currency momentum, Foreign exchange markets, Cross sectional returns, Behavioral finance


    Abstract:

    This study examines whether investor sentiment conditions the profitability of short-term cross-sectional foreign currency momentum strategies. Using weekly spot and forward exchange-rate data for 63 developed and emerging-market currencies over the period 1997–2015, we construct zero-investment winner–loser portfolios based on excess currency returns across alternative formation and holding horizons ranging from one to four weeks. The empirical results provide strong evidence that short-horizon currency momentum exists at weekly frequencies and that its economic significance depends critically on investor sentiment regimes. Specifically, momentum portfolio returns increase monotonically across ranked portfolios. In contrast, the profitability of the long short spread portfolio is concentrated primarily during low-sentiment periods, while it weakens substantially during optimistic-sentiment environments. These findings indicate that investor sentiment operates as a state-dependent conditioning variable governing cross-sectional dispersion in currency returns rather than as a uniform predictor of unconditional momentum profitability. Additional evidence suggests that exchange-rate volatility and macro-financial uncertainty amplify momentum opportunities by widening cross-currency return differentials. Overall, the results highlight the importance of incorporating behavioral indicators into currency allocation strategies and suggest that sentiment-conditioned trading rules improve the timing and risk-adjusted performance of short-term currency momentum portfolios in global foreign exchange markets.


    PDF

Type: Journal

Language: English

Publisher: ya tai jing ji bian ji bu

ISSN: 1000-6052

Email: [email protected]