Demand-Side Fee Flows and Return Predictability on Ethereum

Abstract

We construct a protocol-native valuation signal for Ethereum based on demand-side fees expressed as a share of token supply. The signal measures the log deviation of current fee intensity from its trailing median, a dimensionless ratio denominated entirely in ETH. It predicts subsequent token returns at 10 to 60 day horizons with in-sample R-squared up to 22.8% and expanding-window out-of-sample R-squared of 14.4% at 45 days. The signal retains predictive power after macroeconomic controls, standard crypto risk factors, and momentum controls, and predicts ETH-specific relative returns. Predictability emerges only after the Dencun hard fork (March 2024), which separated execution fees from data availability fees, making the demand signal empirically detectable. Our findings demonstrate that demand-side economic flows are capitalized into token prices in the absence of firms, contracts, or residual cash flow rights, extending valuation logic to rule-based economic systems.