Time-varying relationship of news sentiment, implied volatility and stock returns

I examine the relationship between aggregate news sentiment, S&P 500 index (SPX) returns, and changes in the implied volatility index (VIX). I find a significant negative contemporaneous relationship between changes in VIX and both news sentiment and stock returns. This relationship is asymmetric whereby changes in VIX are larger following negative news and/or stock market declines. Vector autoregression (VAR) analysis of the dynamics and cross-dependencies between variables reveals a strong positive relationship between previous and current period changes in implied volatility and stock returns, while current period and lagged news sentiment has a significant positive (negative) relationship with stock returns (changes in VIX). I develop a simple trading strategy whereby high (low) levels of implied volatility signal attractive opportunities to take short (long) positions in the underlying index, while extremely negative (positive) news sentiment signals opportunities to enter short (long) index positions. The investor fear gauge (VIX) appears to perform better than news sentiment measures in forecasting future returns.