States of psychological anchors and price behavior of Japanese yen futures
Publication date: Available online 9 November 2018
Source: The North American Journal of Economics and Finance
Author(s): Hsiu-Chuan Lee, Yun-Huan Lee, Yang-Cheng Lu, Yu-Chun Wang
This study explores whether the relationship between Japanese yen futures returns and the corresponding equity returns is affected by the states of psychological anchors of the currency and stock markets. This study employs the linear-regression-based tree model (a machine learning method) to account for the framing effect of the anchors. The empirical results of the linear-regression-based tree model show that the currency price behaviors of momentum and reversal, and prediction by equity markets, vary with the anchors. Empirical evidence also indicates that the linear-regression-based tree model outperforms the OLS model based on the estimation results and out-of-sample forecasting. The forecasting performance of the linear-regression-based tree model can be improved along with an increase in the forecasting period.