We are pleased to announce the publication of our new article in the IEEE Xplore digital library, the result of a joint effort between HSDSLab and our industrial partner, CryptoSoft (ICS). The title of the publication is: “Anticipating Crypto Success: An XAI Framework for Early-Stage Token Viability Using Deployment Features.”

This research addresses a key challenge in decentralized finance: can the success of a new cryptocurrency token be predicted at the moment of its deployment, when no market history is yet available.

Our study introduces an eXplainable Artificial Intelligence (XAI) framework that estimates a token's long-term viability based solely on early-stage features. These features include smart contract properties, deployment characteristics, and available metadata. Using a massive dataset of 100,000 ERC-20 tokens from the Ethereum, Binance Smart Chain, and Polygon networks, we demonstrate that models like XGBoost can effectively identify the early signs of future success.

Beyond mere prediction, explainability plays a central role in our framework. SHAP analysis has shown that factors such as code quality, the presence of a visual icon, and the deployment network significantly influence the results.

The research was previously presented by our colleague, Csaba Kiss, at the IEEE ICDLT conference held in November 2025. We would like to thank all co-authors for the fantastic collaboration: Alexy Bounsavath, Csaba Kiss, Tamás Savci, Gábor Hellner, and Dr. Roland Molontay.