Using Options Implied Volatility to Predict Returns

The price of an option (e.g., an option on a stock, for example) can be calculated from its current price, the strike price, the option period and its volatility (e.g., a.k.a. standard deviation).

The price of the stock (when the option is purchased or sold), the strike price and the option period are known factors. Volatility is a more complicated. If volatility is calculated from historical data, it is defined by a sample: a time period over which the volatility is calculated.

The actual options prices that are quoted in the market frequently do not match the historical volatility. Options prices are derived from an estimate of future volatility. This estimate reflects the market maker's, presumably informed, estimate of volatility and the chance that the option may be exercised.

Several papers have speculated that implied volatility can be used as a predictor for the future returns of the underlying asset. This speculation is based on the idea that options implied volatility embodies informed information about the future.

As part of a 1-unit independent study project in graduate school (Computational Finance and Risk Management at the University of Washington) I did a project that investigated whether S&P 500 options implied volatility could be used to predict S&P 500 returns. For the impatient reader: the answer appears to be no. Implied volatility factors have no statistical significance in predicting 1-week ahead returns, at least for the S&P 500.

Reference Journal Article

Although several papers have suggested that options volatility is a predictor of future returns, the paper I choose to work from was Exploiting Option Information in the Equity Market by Guido Baltussen, Bart van der Grient, Wilma de Groot, Erik Hennink and Weili Zhou, Financial Analysts Journal, Vol 68, No. 4 (July/August 2012).

Independent Study Results

My independent study results can be found here.

R Code

The R code that was written to process the options data (from OptionMetrics) can be found here.

Unfortunately the options data that was used in this study is propretary and cannot be redistributed so I can't provide this data. The data was originally obtained from OptionMetrics.

Ian Kaplan
June, 2013
Last updated:

back to Topics in Quantitative Finance