If option volume truly conveys incremental information, it should predict the spread between actual and synthetic stock returns. We test this conjecture and find that both signed and unsigned volumes largely fail to do so, whether around earnings releases, 8-K filings, or other periods. When spread predictability does emerge, it is either statistically weak, economically trivial, or directionally inconsistent with incremental information originating from options. A noisy rational expectations model with informed investors trading in both stock and options motivates our empirical analysis. By focusing on predicting actual stock returns, prior studies may have significantly overstated the informational value of option volumes.