Abstract
As CryptoPunks pioneers the innovation of non-fungible tokens (NFTs) in AI and art, the valuation mechanics of NFTs has become a trending topic. Earlier research identifies the impact of ethics and society on the price prediction of CryptoPunks. Since the booming year of the NFT market in 2021, the discussion of CryptoPunks has propagated on social media. Still, existing literature hasn't considered the social sentiment factors after the historical turning point on NFT valuation. In this paper, we study how sentiments in social media, together with gender and skin tone, contribute to NFT valuations by an empirical analysis of social media, blockchain, and crypto exchange data. We evidence social sentiments as a significant contributor to the price prediction of CryptoPunks. Furthermore, we document structure changes in the valuation mechanics before and after 2021. Although people's attitudes towards Cryptopunks are primarily positive, our findings reflect imbalances in transaction activities and pricing based on gender and skin tone. Our result is consistent and robust, controlling for the rarity of an NFT based on the set of human-readable attributes, including gender and skin tone. Our research contributes to the interdisciplinary study at the intersection of AI, Ethics, and Society, focusing on the ecosystem of decentralized AI or blockchain.
Introduction
While the NFT market has seen astronomical valuations, the underlying "mechanics" of why one digital asset is worth more than another remain opaque. Previous research suggested that "endogenous" factors—like the rarity of an avatar’s traits—and "ethical" factors—like gender and skin tone—influenced prices. Specifically, earlier studies found evidence of racial bias, where CryptoPunks with lighter skin tones sold for higher prices. However, these studies often ignored "exogenous" factors like social media sentiment and failed to account for the massive market transformation that occurred in 2021. This study investigates whether social media "hype" drives price and whether the ethical biases identified in early NFT history persisted after the market went mainstream.
Methodology and Data
The researchers employed a multi-modal approach combining blockchain forensics with natural language processing. The study integrated four distinct data streams: 122,379 general CryptoPunk tweets (analyzed via VADER sentiment analysis), blockchain transaction data from Dune Analytics, Ethereum gas prices from Etherscan, and ETH/USD exchange rates.
To analyze valuation, the authors utilized a Hedonic Regression model. This economic total-value model breaks down the price of a CryptoPunk into constituent parts, including attributes (Gender, Skin Tone), market controls (Rarity, Sales Volume), and a new variable: daily average tweet sentiment scores. Finally, a Structural Break Analysis was performed to compare market behavior across three periods: 2017–2021, 2021–2022, and the full 2017–2022 span.
Sentiment as a Valuation Engine
Social media sentiment is not merely "noise"; the model demonstrates it is a statistically significant contributor to price prediction. When the collective sentiment on Twitter is positive, prices tend to follow. Adding sentiment scores to the hedonic regression improved the model's performance without reducing the significance of other predictors.
A key finding is the "Great Reset" of 2021. This year was not just a price peak but a structural turning point. Post-2021, tweet sentiment became more stable and consistently positive (generally staying between 0 and 0.5) compared to the wild fluctuations of the early years. This stabilization suggests a maturing market where sentiment plays a more predictable role in valuation.
Ethical Imbalances and Structural Shifts
The research documents a major shift in how the market values specific traits. In the early market, female punks often commanded a premium. Post-2021, this reversed, with males predicted to have higher prices. Similarly, the "light skin tone premium" identified in previous literature largely disappeared after 2021. While non-human types (Aliens, Apes, Zombies) still command massive premiums due to extreme rarity, the pricing differences between human skin tones have narrowed or lost statistical significance.
Despite these narrowing price gaps, a stark "attention gap" remains on social media. Discussions are heavily dominated by non-human types and male punks, while female punks and certain skin tones receive significantly less visibility in the digital town square.
Conclusion
This research evidences that social sentiment is a critical factor in NFT valuation, particularly following the market's structural shift in 2021. While some early ethical biases in pricing have faded, the imbalance in social media attention suggests that social biases continue to shape the NFT ecosystem. Future work should expand this analysis to other platforms like Discord and Reddit to further untangle the causal relationship between social hype and digital asset value.
Summary updated