📝 SummarXiv

Abstract

This paper studies how visual traits and market cycles shape prices in NFT markets. Using 94,039 transactions from 26 major generative Ethereum collections, the analysis extracts 196 machine-quantified image features (covering color, composition, palette structure, geometry, texture, and deep learning embeddings), then applies a three-stage filter process to identify stable predictors for hedonic regression. A static mixed-effects model shows that market sentiment and transparent, interpretable image traits have significant and independent pricing power: higher focal saturation, compositional concentration, and curvature are rewarded, while clutter, heavy line work, and dispersed palettes are discounted; deep embeddings add limited incremental value conditional on explicit traits. To assess state dependence, the study estimates a Bayesian dynamic mixed-effects panel with cycle effects and time-varying coefficients for a salient image attribute (Composition Focus - Saturation). Collection-level heterogeneity ("brand premia") is absorbed by random effects. The time-varying coefficients exhibit regime sensitivity, with stronger premia in expansionary phases and weaker or negative loadings in downturns, while grand-mean effects remain small on average. Overall, NFT prices reflect both observable digital product characteristics and market regimes, and the framework offers a cycle-aware tool for asset pricing, platform strategy, and market design in digital art markets.