Drawing upon more than 12 million observations over the period from 1996 to 2020, we find that allowing for nonlinearities significantly increases the out-of-sample performance of option and stock characteristics in predicting future option returns. The nonlinear machine learning models generate statistically and economically sizable profits in the long-short portfolios of equity options even after accounting for transaction costs. Although option-based characteristics are the most important standalone predictors, stock-based measures offer substantial incremental predictive power when considered alongside option-based characteristics. Finally, we provide compelling evidence that option return predictability is driven by informational frictions and option mispricing.

Contrary to the theoretical principle that higher risk is compensated with higher expected return, the literature shows that low-risk stocks outperform high-risk stocks. Using a large-scale household dataset, we provide an explanation for this puzzling result that the anomalous negative risk-return relation is only confined to those stocks predominantly held by rich households, whereas the anomaly disappears for stocks held by non-rich households and institutional investors. We find that social status concern of rich households and the induced lottery preference explain wealthy investors’ demand for high-risk stocks, leading to overpricing and low future returns for such stocks.