Does relative demand for call and put options on individual stocks, as measured by average difference in implied volatilities of at-the-money calls and puts (aggregate implied volatility spread), predict stock market returns? In their September 2017 paper entitled “Aggregate Implied Volatility Spread and Stock Market Returns”, Bing Han and Gang Li test aggregate implied volatility spread as a U.S. stock market return predictor. They focus on monthly measurements, but test the daily series in robustness test. They calculate monthly implied volatility spread for each stock with at least 12 daily at-the-money call and put option prices during the month as an average over the last five trading days. They then eliminate outliers by excluding the top and bottom 0.1% of all stock implied volatility spreads before averaging across stocks to calculate aggregate implied volatility spread. They compare the predictive power of aggregate implied volatility spread to those of 22 other predictors from prior research. Using daily at-the-money call and put implied volatilities for U.S. stocks, data for other U.S. stock market predictors and U.S. stock market returns during January 1996 through December 2015, *they find that:*

- Aggregate implied volatility spread is negative about 80% of the time, indicating a larger premium for put options, with average value -0.8%.
- Based on correlations, aggregate implied volatility spread is somewhat like variance risk premium (0.35) and market variance (0.48), and somewhat opposite TED spread (-0.34) and default yield spread (-0.34).
- Aggregate implied volatility spread relates positively to future U.S. stock returns, both in-sample and out-of-sample, up to six months ahead.
- At monthly and quarterly horizons, aggregate implied volatility explains 5.5% and 12.8% (5.8% and 12.4%) of variations in future stock market returns in-sample (out-of-sample), respectively.
- Predictive power concentrates in morning (9:00AM-12:00PM) stock market returns, consistent with morning releases of most U.S. economic news.
- Predictive power is robust to alternative metrics, across subperiods and after controlling for other known stock market return predictors.
- For out-of-sample tests during 2002-2015, aggregate implied volatility spread is a significantly better predictor of stock market return than other known predictors.

- In contrast, the implied volatility spread of S&P 500 index options exhibits no power to predict U.S. stock market returns at monthly and longer horizons.
- A mean-variance optimization investor who exploits aggregate implied volatility spread return forecasts to shift allocations monthly between the U.S. stock market and U.S. Treasury bills achieves a 0.65 gross annual Sharpe ratio during 2002-2015, compared to 0.43 for the market portfolio. The best other predictors exploited similarly achieve gross annual Sharpe ratios 0.59 (short interest index) and 0.49 (variance risk premium).
- The predictive power of aggregate implied volatility spread is consistent with widespread informed trading in equity options.

In summary, *evidence indicates that the average spread between at-the-money call and put option implied volatilities for U.S. stocks relates positively and exploitably to future U.S. stock market returns.*

Cautions regarding findings include:

- Reported returns for the portfolio test are gross, not net. Accounting for any costs of monthly reallocations would reduce returns and Sharpe ratio.
- Testing more and more potential predictors on the same data (U.S. stock market returns), including tests not reported due to lack of success, increases the likelihood of discovering a false positive (a variable that is lucky in the sample).
- Calculation of aggregate implied volatility spread involves data costs and is beyond the capability of most investors, who would bear fees for delegating the activity to an advisor/investment manager.

See also “Aggregate Stock Option Put-Call Ratio as Market Return Predictor”.