Summary queued
This paper remains searchable while structured annotation is pending.
ECON CURRENT excludes papers without a reviewed, schema-valid annotation from editorial selection. Public metadata remains available below.
Public abstract
We construct a novel stock-level measure of volatility disagreement as the cross-sectional dispersion of realized variance forecasts built from heterogeneous information sets and machine learning models. Sorting single-name delta-hedged straddles on this measure yields a long-short return of −5.14% per month, robust to known option return predictors and not subsumed by disagreement about first moments or the variance risk premium. A one-standard-deviation increase in volatility disagreement is associated with a 30% rise in option position opening. The cross-sectional patterns of our disagreement measure align with recent theoretical advances of beliefs and pricing of variance claims. Evidence on attention, ownership, and arbitrage costs is more consistent with mispricing than risk compensation.