Unpacking the black box of ICO white papers: A topic modeling approach

Abstract

We apply a novel topic modeling method to map Initial Coin Offerings’ (ICOs’) white paper thematic content to analyze its information value to investors. Using a sentence-based topic modeling algorithm, we determine and empirically quantify 30 topics in an extensive collection of 5,210 ICO white papers between 2015 and 2021. We find that the algorithm produces a semantically meaningful set of topics, which significantly improves the model performance in identifying successful projects. The most value-relevant topics concern the technical features of the ICO. However, we find that white paper's informativeness substantially diminishes after the token is listed. Moreover, we show that credibility-enhancing mechanisms (i.e., regulations and ICO analysts) reinforce the information value of ICO white papers. Overall, our results suggest that the topics discussed in white papers and the attention devoted to each topic are useful ICO performance indicators.