I examine whether complex disclosures of forward-looking, value-relevant information that is not in financial statements influence investor disagreement at subsequent earnings announcements. Conventional wisdom suggests that more extensive disclosure decreases information asymmetry among investors, which should mitigate disagreement in their interpretations of earnings. However, recent theories suggest that when disclosures contain more complex information that is difficult to interpret, investors are more likely to interpret the information differently. These differences can result in more disagreement among investors not only at the disclosure date, but also at the subsequent earnings announcement as they incorporate the pre-announcement differences in beliefs in their valuations of earnings. I examine my research question using new product announcements (NPAs), which provide information about new product introductions, such as the expected product release date, prices, product attributes, and other technological details. Using machine learning algorithms to assess the narrative content of NPAs, I predict and find that more complex components of NPAs are associated with greater investor disagreement at the NPA and subsequent earnings announcements. My findings suggest that efforts to improve the valuation of intangible assets such as product innovation through more detailed and complex disclosures may actually lead to greater investor disagreement over the valuation of firm earnings.