This project studies how NFT prices actually form, using transaction-level data from the Bored Ape Yacht Club collection pulled from OpenSea and Etherscan. It combines econometric modeling with machine learning to disentangle three competing explanations for value: the token’s intrinsic traits (rarity, visual attributes), the timing of a sale, and social signaling from buyer and seller reputations.

The headline finding is that trait rarity dominates pricing early on, but that effect decays over time as the reputations of the accounts involved become the stronger predictor of price. The data also shows trading patterns consistent with strategic, reputation-building behavior among high-volume wallets — evidence that NFT markets behave less like static collectibles markets and more like ongoing reputation economies.

Abstract

Prior work on NFT valuation has emphasized intrinsic traits such as rarity and visual features. Using Bored Ape Yacht Club data collected from OpenSea and Etherscan, this study shows that while traits dominate initial valuations, their effect declines over time as buyer and seller reputations emerge as stronger determinants of price. The analysis also identifies trading patterns suggesting strategic, reputation-driven behavior among high-volume accounts, indicating that NFT valuation results from the combined influence of design characteristics, participant reputation, and trading behavior.