Blockchain technology has gained tremendous importance in recent years, paving the way for cryptocurrencies and crowdfunding via Initial Coin Offerings (ICOs). However, the possibility to raise vast amounts of money in an anonymized and unregulated market also attracts criminals who cause substantial damage to investors. In the present study, we build a predictive model to identify ICO fraud. To this end, we collect data on ICOs in the form of whitepapers and apply natural language processing techniques as well as state-of-the-art machine learning algorithms. We train our model on features that are derived from established fraud theory. Our model indicates a high predictive power with more than 80 percent accuracy on the test set. This study contributes to the field of investment scam detection from both a theoretical and a practical perspective. On the one hand, our study shows that the theorygrounded linguistic features prove to be predominantly relevant for fraud prediction. However, the results do not always coincide with what could be expected from theory. On the other hand, our predictive model may support potential ICO funders in their investment decision. Early fraud detection will help investors to reduce the investment risk and thus not lose their funds.
Dürr, Alexander; Griebel, Matthias; Welsch, Giacomo; and Thiesse, Frédéric, "PREDICTING FRAUDULENT INITIAL COIN OFFERINGS USING INFORMATION EXTRACTED FROM WHITEPAPERS" (2020). In Proceedings of the 28th European Conference on Information Systems (ECIS), An Online AIS Conference, June 15-17, 2020.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.