This paper examines whether abnormal trading helps or hurts price discovery. I use options data and study the relation between realized and model-free, as well as Black-Scholes, implied volatilities across high and low option volume states. I also disaggregate option volume into speculative and hedging components and test how abnormally high trading volumes accompanied by an abnormally high speculative component (or by an abnormally high hedging component) affect price discovery. The results reveal that when option volume is high, implied volatilities are more informative. This is consistent with the hypothesis that abnormal volume reflects shocks to informed traders' demand. The result remains robust to alternative weighting schemes for option volume based on option spread and vega weights, which indicate the type of contracts informed traders are likely to prefer. As an important econometric robustness check, I employ the Kalman filter in order to directly address the latent feature of the volatility risk premium in the implied - realized volatility relation.