Valuation of crypto assets on blockchain with deep learning approach
Xi Zhou, Yin Pang, Esther Ying Yang, Jing Rong Goh, Shaun Shuxun Wang · Quantitative Finance and Economics · 2025
With the rapid expansion of the blockchain ecosystem, crypto asset valuation has become an essential area of study for investors and institutions. Here, we introduce a deep learning framework that was designed to predict the value index of crypto assets by integrating intrinsic value variables and decomposing market prices into value and sentiment components. The Crypto Asset Value-indexing Model (CAVM) was applied to Ethereum's cryptocurrency ETH to demonstrate its effectiveness. Four econometric tests were conducted to verify the informativeness, predictiveness, and reasonability of the generated value indices, as well as the efficiency of price decomposition. Our findings suggested that the value index can serve as a reliable proxy for the intrinsic value of crypto assets, offering a benchmark for investment decisions, consumption, financial reporting, and potential tax implications. Additionally, this research contributes to the literature on asset valuation by proposing a novel method that applies deep learning techniques to intangible assets traded in secondary markets. By utilizing the end-to-end nature and directed acyclic graph structure inherent to deep learning models, we enhance the modeling process with customized loss functions and regularization mechanisms.