Forecasting stock returns: the role of VIX-based upper and lower shadow of Japanese candlestick

This paper proposes a new predictor by calculating the difference between the Japanese candlestick’s upper and lower shadows (ULD) constructed from CBOE volatility index (VIX) data. ULD is a powerful predictor for future stock returns, and higher ULD leads to the subsequent decline of stock returns. Our results show that our new predictor generates R^2 values of up to 2.531% and 3.988% in-sample and out-of-sample, respectively; these values are much larger than the previous fundamental predictors. Moreover, the predictive information contained in ULD can help mean–variance investors achieve certainty equivalent return gains of as high as 327.1 basis points. Finally, the extension analysis and robustness tests indicate that recession is the primary cause of return predictability; our results are robust under different settings.