The Anatomy of a Cryptocurrency Pump-and-Dump Scheme (1811.10109v2)

Abstract: While pump-and-dump schemes have attracted the attention of cryptocurrency observers and regulators alike, this paper represents the first detailed empirical query of pump-and-dump activities in cryptocurrency markets. We present a case study of a recent pump-and-dump event, investigate 412 pump-and-dump activities organized in Telegram channels from June 17, 2018 to February 26, 2019, and discover patterns in crypto-markets associated with pump-and-dump schemes. We then build a model that predicts the pump likelihood of all coins listed in a crypto-exchange prior to a pump. The model exhibits high precision as well as robustness, and can be used to create a simple, yet very effective trading strategy, which we empirically demonstrate can generate a return as high as 60% on small retail investments within a span of two and half months. The study provides a proof of concept for strategic crypto-trading and sheds light on the application of machine learning for crime detection.

• The paper presents a comprehensive empirical analysis of 412 pump-and-dump events, revealing that smaller market cap coins are especially vulnerable to manipulation.
• The methodology employs a random forest model with an AUC over 0.90, using short-term market signals like hourly returns and trading volumes for prediction.
• Implications include enhanced trading strategies and regulatory tools to detect and mitigate artificial trading volumes and insider manipulations in crypto markets.

An Empirical Analysis of Cryptocurrency Pump-and-Dump Schemes

The paper "The Anatomy of a Cryptocurrency Pump-and-Dump Scheme" by Jiahua Xu and Benjamin Livshits offers a comprehensive empirical investigation into pump-and-dump schemes within the cryptocurrency markets, with a specific focus on the role of social media platforms like Telegram and exchanges such as Cryptopia, Yobit, Binance, and Bittrex. This research stands out by providing an analytical framework and a predictive model for understanding and potentially anticipating pump-and-dump activities.

Overview and Empirical Findings

The authors leverage an extensive dataset, comprising 412 pump-and-dump events from June 2018 to February 2019, organized primarily through Telegram channels. They conduct a detailed analysis of these events, uncovering crucial patterns that characterize such schemes.

One of the key empirical findings is the role of coin market capitalization in the success of pump-and-dump operations. Smaller market cap coins tend to be targeted more frequently due to their susceptibility to price manipulation, akin to microcap stock manipulations in traditional financial markets. This aligns with previous work, such as Hamrick et al. (2018), supporting the notion that smaller market cap assets present more lucrative targets for market manipulators.

The paper identifies that pump-and-dump activities generate significant artificial trading volumes—amounting to millions of USD monthly—indicative of the broad market impact these schemes can have. Abnormal price movements in the hours preceding pump events suggest potential insider trading by organizers, highlighting the informational asymmetries exploited in these schemes.

Machine Learning Model for Prediction

Importantly, the paper contributes to the field by developing a random forest-based machine learning model that predicts which coins might be targeted for future pump-and-dump activities. This model achieves high predictive accuracy, with an AUC (Area Under Curve) exceeding 0.90, signifying a robust ability to identify potential targets by analyzing pre-event market signals. Key predictive features include short-term market indicators like hourly returns and trading volumes, rather than broader historical data, demonstrating the models' reliance on immediate market movements preceding an event.

The model's performance suggests practical applications, providing cryptocurrency traders and regulators with a tool for anticipating and potentially mitigating the impacts of pump-and-dump schemes. By applying the model in a simulated trading strategy, the authors illustrate potential returns of up to 60% over a period of two and a half months, underscoring the model's utility in developing strategic trading approaches.

Implications and Future Research

This research has significant implications for market participants and regulators. For traders, the predictive model offers a strategic advantage, enabling more informed trading decisions. For regulators, the paper presents an opportunity to integrate machine learning techniques for real-time detection and prevention of market abuses, contributing to fairer market environments.

Theoretically, the paper advances understanding of market manipulation within the cryptocurrency domain, adding empirical weight to the argument that smaller-cap coins are more vulnerable. Future research could explore the integration of additional data sources, such as order books and social media sentiment analysis, to enhance predictive capabilities further. Additionally, expanding the analysis to include other emerging exchanges and platforms could provide a more comprehensive view of the landscape of cryptocurrency fraud.

Overall, the paper represents a significant step forward in the empirical paper of cryptocurrency markets, offering both detailed insights into pump-and-dump mechanisms and a practical tool for prediction and strategic response.