There are many instances where people are willing to pay a premium for rare items or unique experiences. It's no surprise that rare NFTs sell for higher prices, but it's also important to quantify the value that rarity implies.
From a business perspective, many assets do not require any form of restricted supply. The value of rarity will affect the sum of consumer demand and market-level demand, hence artificially designed rarity and quantity will produce different effects. Rarity value is economically rooted in the concept that resources are finite and demand is infinite, which is reflected through price. The value of blue-chip NFTs is divided into several major aspects.
Firstly, the scarcity of the item—the issued quantity per collection.
Secondly, the traits compared to other NFTs in the total collection. In addition, rarity is closely related to the notion of conspicuous "proprietary" consumption, that is, consumers' access to communities through the purchase of NFTs also represent their attitude and identity in some cases.
Thirdly, it is the usage value, or premium, that comes with the scarcity of NFT. This includes the enjoyment of trading NFTs and their extensive application in GameFi.
Lastly, the difficulty of acquisition or the value of time. Scarcity mirrors the cost of community maintenance, which can be seen through community passes. A higher cost of maintenance shows on the price and thus forming a closed loop.
So, is the rarity the major factor that drives price change? Since January, the market demand for NFT has dropped, but blue-chip projects have continued to rise in price due to their scarcity, thus reflecting positive consumer expectations for top assets.
However, could there be a quantitative system that explores the intrinsic correlation between NFT price and rarity so as to detect a pattern? With the help of on-chain records of NFT transactions, we are able to assess the impact of NFT rarity on its price.
Note: We divided the NFTs into four groups according to their rarity, with "x" denoting the rarity of the NFTs.
90 ≤ x ≤ 100: Legendary
70 ≤ x < 90: Rare
40 ≤ x < 70: Classic
0 ≤ x < 40: Normal
The price of different NFTs is affected by rarity in different ways. The chart below shows the rarity of the top ten NFT collectibles as well as their total time in circulation on the secondary market.
In most scenarios, the quest for rarity exists, but the individual value of an NFT other than its collectible nature will largely influence the magnitude of rarity's impact on price. The weakened influence of rarity tends to give way to aesthetic value, community consensus, and other factors that are more implicit than the rarity.
CryptoPunks is marketed more towards an OG NFT project, and its value is pretty much self-evident. Rarity is certainly an extremely important flavor enhancer in the collecting experience. As for BAYC, although the project is also a classic and relatively early blue-chip NFT, its community and business system have constructed some of the NFTs' additional values and thus diluted the impact of rarity on its price. Compared to CryptoPunks, its percentage of legendary rarities in the top 10 most expensive NFTs has decreased significantly.
Therefore, the rarest NFT is generally not the expensive one.
By comparing the data of two NFTs with different rarity levels, namely Normal (bottom 40%) and Legendary (top 10%), we observed that rarity affects different NFTs to varying degrees. For PFP projects, in general, the higher the rarity, the higher the price, and NFTs with a 1/1 top trait tend to have higher selling prices. As we can see in the chart below, the huge difference in rarity between Cryptopunks and Doodles has caused a spike in the average price of NFTs. As for BAYC, it is clear that rarity does not contribute to a distinct price tiering and does not represent the core value of the particular NFT.
After comparing the price tiering gap between the highest and lowest rarity, there is another question that users often encounter: how should one choose among the other 90% of NFTs if they can't afford the top 10%? We removed the top 10% of high rarity NFTs and further analyzed NFTs that are relatively low in rarity. We collected NFTs with rarity rankings between 2000 and 4000 as the data set (listed as Medium NFT) and compared their prices with NFTs ranked below 4000 (listed as Bottom NFT). Many would assume that Medium NFT holds rarer NFTs than Bottom NFT and should be priced higher. Yet, the results suggest differently. The following chart shows the average sale price (USD) for both types of NFTs.
We note that there is a clear dividing line between NFTs with high rarity rankings and those with low rankings.
To explore the effects of high rarity, we included NFTs with Legendary rarity and selected NFTs ranked within the 2000 range (called Top NFT). The following chart shows the average price (USD) of each group of NFTs.
It is easy to see that even though Medium NFTs have a higher status in the market than Bottom NFTs, there is no significant difference in their market valuation. On the other hand, the value of Top NFTs remains higher than that of Medium NFTs. It is clear that NFT prices are influenced by the "head effect".
Trading behaviors in the market include "floor sweeping" and "head collecting". It is rare to see buyers collecting NFTs in the middle range. Even for some GameFi projects that correlate rarity with farming speed, collecting NFTs of medium rarity does not appear to be very economical compared to the other two strategies. Therefore, in the psychological and economical sense, the effect of rarity on the price for mid-range NFTs is somewhat "dysfunctional".
Quantitative analysis of large-scale NFT transactions helps us better understand the causes of outliers. First, we analyzed the highest prices of a collection using Z-score and standard deviation. As we measure the deviation of the data from the average of the dataset in which it is located, we also assess the diversity of NFT collection prices to understand how sellers drive the market. Azuki's selling prices show that most of its NFTs sold at prices below the average selling price, and some of its Z-scores range from 3 to 7. How many of these outliers are caused by rarity? The results of Azuki's rarity distribution chart show that most of the outliers have a Normal rarity level.
To further analyze the correlation coefficient, we introduce some well-known statistical analysis methods in finance. Pearson Correlation has been widely used for correlation analysis and the following is the formula for Pearson Correlation Coefficient (PCC).
x denotes rarity and y denotes the latest selling price. Pearson Correlation (PC) only holds when each dataset is a normal distribution. NFT data are usually in non-normal distribution. Therefore, in order to evaluate the distributions of the two data sets (NFT rarity and latest prices), we collected data on rarity and the latest prices on the 6 collectibles samples group.
It is easy to see that although certain collections, such as BAYC, show a normal distribution, the data distribution of the NFT collection is still non-normal in terms of prices. In the case where price data is non-normal, we employ the QQ-plot method to test if the data conform to normal distribution. This is done by selecting a data sample, i.e. NFT prices, and comparing it with normal distribution. The red line on the graph represents the data conforming to normal distribution while the dots represent real data.
We found that a portion of the data set is normally distributed, but we have to take into account all the outliers in the collection when choosing our analysis method. In general, NFT prices tend to be non-normally distributed, and some studies confirm that it is also this tendency that leads to errors in the estimates of the Pearson correlation coefficient.
The non-normal distribution causes the correlation coefficient to be inflated by +0.14. Although many standard error elimination mechanisms can be used to solve this problem, we have chosen a more robust approach: Spearman correlation, which is a relatively conservative estimate of the correlation between rarity and price. The following is a simplified formulation of Spearman correlation.
Spearman Correlation The result of this equation is a value between -1 and 1, indicating that the relationship is completely positive or completely negative. The closer this value is to 0, the more the correlation shows a non-negative correlation. Spearman correlation varies greatly between collections. According to the calculations, the correlation between the rarity of certain collections and the price is small, meaning that there are other variables, whereas the rarity of certain collections is significantly correlated with their price. The figure below shows the difference between the results of the Pearson correlation and Spearman correlation.
The results of this study show that Spearman's method produces fewer mistakes while Pearson's method produces over-or underestimation for all collections due to the presence of non-normal distributions.
The final results show that, among these six blue-chip projects, Doodles has the strongest rarity-price correlation. On the contrary, BAYC clearly has other variables that affect its price. Apparently, people value the additional aspects of BAYC as well as its rare monkeys.
Like the Panini NBA cards, price tiering occurs when the number of players in a category of collectibles is large enough. Panini's scoring system includes centering, corners, edges, and surfaces. Similar to this idea, NFT prices are determined not just by rarity, but more by their community. For example, StartCatchers have different dynamic traits, and the community defines 3 dynamic traits > 2 > 1 > static. Furthermore, the value of the mfers project lies in the cultural consensus behind it. Holders resonate with the NFT images as they think these images reflect a part of them. The influence of rarity on price is greatly diluted and replaced by aesthetic resonance and community consensus. In addition, NFT prices are also related to the celebrity effect as some traits associated with celebrities sell for higher-than-average prices.
Thus, the community's ability to create more strategies and approaches will make it easier to promote and increase circulation, especially for the PFP project. On the flip side, the value of NFT is determined and influenced by a variety of factors. By collaborating with major celebrities, PFP projects can implicitly increase the price of NFTs by redefining rarity while expanding their own influence. Meanwhile, some NFT Pricing and Valuation tools need to reconstruct their model with rarity, this factor.
Through constructing a data analysis mechanism, we explored the correlation between NFT rarity and price, quantitatively analyzed NFT pricing power, and reassessed the impact of rarity on price. At the same time, we further analyzed the characteristics of different collections based on dynamic factors such as their secondary market circulation time, project activity, and variability for the purpose of helping investors to better plan their strategies and expectations of returns in the NFT market.