Asset pricing bubbles are highly emotional market episodes. Despite this, investor emotions are not part of traditional bubble models. We measure the powerful affects influencing investor decisions during speculative market bubbles directly employing textual analysis of media narratives and domain-specific emotion keyword dictionaries and show how understanding investor emotional dynamics helps explain market behavior. Specifically, we focus on the two Chinese stock market bubbles of 2005–2008 and 2014–2016; there is no evidence of investor learning from experience. Despite Chinese media being censored we show it still has strong explanatory power although the independent English language media can provide an additional perspective. Deeper emotions dominate more superficial feelings in information content.