Trading Options with Uncertainty Risk around Earnings Announcements

Abstract

Quarterly earnings announcements release a substantial amount of financial information about a firm all at once, often inducing significant, discontinuous jumps in stock prices. Traditional option pricing models, such as the Black-Scholes model and stochastic volatility models, fail to model these jumps. In this thesis, we aim to estimate the earnings volatility as priced into existing options contracts. To do so, we use a modified Black-Scholes model and a modified Heston model that incorporates jumps at earnings. Since both Black-Scholes and Heston have differing underlying models for the stock price, we obtain two distinct estimates for earnings volatility. These estimates are observed to be similar, although the estimate derived from the Heston model is marginally higher. We then attempt to explain the estimates of the earnings volatility using its lags, market volatility, and quarter information, among other features. We observe that the previous two earnings volatilities are most predictive of future earnings volatility. We then build more sophisticated prediction-focused models and observe that a LASSO model exhibits the best out-of-sample error. These predictive models allow us to construct four distinct trading strategies based on the predicted earnings volatility, which are evaluated against several benchmarks and prove to be profitable in a back-test. This work shows that simple heuristic-based trading strategies can exploit discrepancies in the valuation of volatility in options contracts around earnings announcements.