Heterogeneous rarity patterns drive price dynamics in NFT collections

TL;DR

It is shown that, on average, rarer NFTs sell for higher prices, are traded less frequently, guarantee higher returns on investment, and are less risky, i.e., less prone to yield negative returns.

Abstract

We quantify Non Fungible Token (NFT) rarity and investigate how it impacts market behaviour by analysing a dataset of 3.7M transactions collected between January 2018 and June 2022, involving 1.4M NFTs distributed across 410 collections. First, we consider the rarity of an NFT based on the set of human-readable attributes it possesses and show that most collections present heterogeneous rarity patterns, with few rare NFTs and a large number of more common ones. Then, we analyze market performance and show that, on average, rarer NFTs: (i) sell for higher prices, (ii) are traded less frequently, (iii) guarantee higher returns on investment (ROIs), and (iv) are less risky, i.e., less prone to yield negative returns. We anticipate that these findings will be of interest to researchers as well as NFT creators, collectors, and traders.

Figures (21)

• Figure 1: Illustration of rarity in a collection. Demonstration of the visual difference between rare and common NFTs using the example of CryptoPunks. CryptoPunk #2547 (on the right) is the least rare, as it has traits that appear frequently in the collection (i.e., the bandanna and the earring). CryptoPunk #8348 is the rarest in the collection, mostly since it is the only one with seven non-null attributes. Rarity scores are not normalised. After normalisation, the total rarity score for punk #2547 is zero, while the one for #8348 is 100 (min and max of the collection, respectively). In the bottom left corner, we show, as an example, the rarity score of traits associated to the "Facial Hair" attribute.
• Figure 1: Collectible NFTs minted over time. Distribution of the collectible NFTs considered in this analysis minted over time.
• Figure 2: Characteristics of collections. The probability distribution of a) the total number of NFTs within the collection; b) the total number of attributes in the collection; c) the total number of traits in the collection.
• Figure 2: Distribution of the log-normal distribution characteristic parameter $\mu$. Distribution of the log-normal distribution parameter $\mu$ (blue line), and its average value across collections (red dashed line). The log-normal distribution $\ln(X)\sim {\mathcal{N}}(\mu, \sigma^{2})$ captures the distribution of rarity for $90\%$ of collections.
• Figure 3: The collectible market boom. 30-day rolling average of the daily number of primary sales (blue line), secondary sales (orange line) and new buyers (i.e., new wallet addresses buying an NFT for the first time, green line). Inset: 30-day rolling average of the daily trade volume (in USD).