Where does value come from: Is there a correlation between blue-chip NFT rarity and price?

Author: NFTGo

In many cases, people are willing to pay a premium for rare items or unique experiences. As the saying goes, "rarity increases value," but how do we quantify the rarity of digital assets and their corresponding value in the world of NFTs?

From a purely commercial perspective, many assets do not require any form of supply limitation, but the value of rarity will influence consumer demand and the overall market demand. Artificially designing rarity and quantity will produce different effects. The economic roots of rarity value lie in the concept that resources are limited while demand is infinite, and price serves as a signal of resource scarcity.

The rarity value of blue-chip NFTs can be measured by the following four points: first, the rarity of the project itself—each collection issues a specific number, mostly around 10,000.

The second is the rarity value compared to other NFTs within the total collection. Rarity is closely related to the concept of "exclusive" consumption—similarity and aesthetic perception between the NFT and its owner can, in some cases, represent an emotional value of attitude and identity. For example, the Azuki held by Jay Chou and its inherent homogeneity.

The third is the utility value brought by the scarcity effect of NFTs, or the premium, including the enjoyment of buying and selling NFTs, and various applications and extensions in the new fields of GameFi and the metaverse.

The fourth is the difficulty of acquisition or time value. The value of scarcity also reflects the community maintenance costs. For instance, the operational costs of some community passes will be reflected in their prices, forming a closed-loop flywheel.

So, is rarity the dominant influencing factor for price? Since January, although the overall trading volume of NFTs in the market has declined, blue-chip projects that were previously agreed upon in the market consensus have still seen price increases due to their rarity, reflecting consumer optimism for top-tier assets.

We can quantify the intrinsic correlation between NFT prices and rarity through data, thereby exploring possible patterns and trends. We selected six distinct blue-chip projects to assess the impact of NFT rarity on their prices.

TL; DR

Holders of rare NFTs have more pricing power than ordinary NFT holders by more than ten times.

Contrary to the common perception that "rare NFTs are more expensive," the impact of rarity on price is not entirely positively correlated.

The influence of rarity may sometimes yield to aesthetic value, community consensus, and other more implicit factors relative to rarity.

Even though Medium NFTs rank higher in rarity than Bottom NFTs in the market, the market price stratification effect between the two is not obvious. The value of Top NFTs far exceeds that of Medium NFTs.

Among the listed six projects, Doodles has the strongest correlation between rarity and price, while BAYC clearly has other variables affecting its price.

Note: Based on different rarity levels, we divided NFTs into four groups, where x represents the rarity level of the NFT:

90 ≤ x ≤ 100: Legendary

70 ≤ x < 90: Rare

40 ≤ x < 70: Classic

0 ≤ x < 40: Normal

Is the highest-priced NFT the one with the highest rarity?

The prices of different NFTs are influenced by rarity to varying degrees. The following chart shows the rarity levels of the top ten priced NFTs among various collections, as well as their total circulation time in the secondary market.

Rarity distribution of the top ten priced NFTs in various blue-chip collections; Source: NFTGo.io

In most scenarios, many people will use various tools to check rarity. However, aside from collectible value, other values of NFTs can significantly dilute the impact of rarity on price. The influence of rarity may sometimes yield to aesthetic value, community consensus, and other more implicit factors.

CryptoPunks is more positioned as an OG NFT project, and its collectible value as the first NFT project is self-evident, while rarity is undoubtedly an extremely important enhancer in the collectible experience. For BAYC, although it is also a very classic and relatively early blue-chip NFT, the community and business system behind the project support some of the other values of the NFT, thereby dispersing the influence of rarity on its price. The proportion of Legendary rarity among the top ten priced NFTs has significantly decreased compared to CryptoPunks.

Therefore, from the perspective of different project positioning, the highest-priced NFT is not necessarily the one with the highest rarity.

Does rarity affect price stratification?

By comparing the data of two NFTs with different rarity levels—the most common Normal and the rarest Legendary (the bottom 40% and the top 10%, respectively)—we can see that the impact of rarity varies among different collectibles. For PFP projects, generally speaking, the higher the rarity, the higher the price, and NFTs with 1/1 top traits often command higher prices.

From the chart below, we can see that the significant difference in rarity among Cryptopunks and Doodles projects leads to a surge in the average price of NFTs. However, for BAYC, it is evident that rarity does not exhibit a clear price stratification; rarity does not represent or reflect the core value of the NFT itself.

Average NFT prices based on rarity rankings (in USD); Source: NFTGo.io

Head and Long Tail Effects

After comparing the price stratification gap between the highest and lowest rarity, another question users often encounter is: if they cannot afford the high-priced NFTs in the top 10% rarity, how should they choose NFTs in the remaining 90% rarity range? We removed the high rarity NFTs in the top 10% and further analyzed the relatively lower rarity NFTs.

By collecting NFTs ranked between 2000 and 4000 in rarity as a dataset (referred to as Medium NFTs) and comparing their prices with NFTs ranked below 4000 (referred to as Bottom NFTs), many people would assume that Medium NFTs contain rarer NFTs than Bottom NFTs, and thus their prices should be higher. Is this really the case? The following chart shows the average prices of the two types of NFTs (in USD).

Average prices of NFTs with different rarity rankings (in USD); Source: NFTGo.io

We can see a clear dividing line between NFTs with higher rarity rankings and those with lower rarity rankings. To study the effect of high rarity, we included NFTs with Legendary rarity and selected NFTs ranked within the top 2000 (referred to as Top NFTs). The following chart shows the average prices of each group of NFTs (in USD):

Source: NFTGo.io

It is evident that even though Medium NFTs rank higher than Bottom NFTs in the market, the market price stratification effect between the two is not obvious. On the other hand, although Medium NFTs are relatively rare in the market, the value of Top NFTs still far exceeds that of Medium NFTs. Clearly, a "head effect" influences the prices of NFTs.

There are various trading behaviors in the market. We often hear about floor sweeping, collecting top traits, and 1/1 traits, but there is little discussion about collecting mid-tier NFTs. Even for some GameFi projects related to rarity and gold mining, collecting medium rarity NFTs seems less economical compared to the other two strategies.

Therefore, due to psychological and economic reasons, the influence of rarity on price exhibits a certain "failure" in the mid-tier.

Which group of NFTs has the strongest correlation?

To further verify the correlation between rarity and price, we also used z-score normalization, standard deviation, and Pearson correlation coefficient to analyze the prices of collections. This measures the extent to which data deviates from the average of its dataset while assessing the diversity of NFT collectible prices, thereby understanding how sellers drive the market.

Pearson Correlation

Here, x represents rarity, and y represents the latest price. The Pearson correlation only holds when each dataset is normally distributed, and NFT data is not even approximately normal; it is usually more non-normal. To assess the distribution of the two datasets (NFT rarity and latest price), we compiled the rarity and price of NFT samples from six collections, as shown below:

Distribution of NFT rarity and prices; Source: NFTGo.io

From the chart above, it is clear that although some collectibles, such as BAYC, exhibit a normal distribution, the price distribution of NFT collectibles remains non-normal, indicating that the impact of rarity on price is minimal. In the case of a non-normal price dataset, we used the QQ-plot method to test whether the data conforms to a normal distribution. We compared NFT prices with a normal distribution—the red line on the chart represents data that conforms to a normal distribution, while the points represent actual data.

QQ-plot of NFT prices; Source: NFTGo.io

We observe that a portion of the dataset conforms to a normal distribution, but when selecting analysis methods, we must consider all outliers in the collectibles. Generally, NFT prices tend to be non-normally distributed, and some studies confirm that this tendency leads to errors in estimating the Pearson correlation coefficient.

The non-normal distribution inflated the correlation coefficient by +0.14, and we used a more robust method for verification—Spearman correlation, which provides a relatively conservative estimate of the correlation between rarity and price. Below is the simplified formula for Spearman correlation.

Spearman Correlation

The result of this formula is a value between -1 and 1, indicating a completely positive or completely negative relationship; the closer the value is to 0, the more the correlation appears non-negative. The Spearman correlation varies significantly among different collectibles. According to the calculation results, some collectibles have a very small correlation between rarity and price, indicating that other variables exist, while some collectibles have a significant correlation between rarity and price. The chart below shows the differences between the results of Pearson correlation and Spearman correlation analysis.

Differences between Pearson correlation and Spearman correlation; Source: NFTGo.io

The research results indicate that the errors produced by the Spearman algorithm are smaller, while using Pearson correlation leads to overestimation or underestimation of the results for all collectibles due to the presence of non-normal distribution.

The final statistical results show that among the six projects studied, Doodles has the strongest correlation between rarity and price, while BAYC clearly has other variables affecting its price. Perhaps this is why BAYC has become one of the most successful NFT projects in history; beyond a rare ape, people also value the other more valuable aspects that BAYC brings.

Is price defined by the community or rarity?

Just like Panini NBA trading cards and rare game cards, when the player community for a type of collectible is large enough, price stratification will occur. NFTs are similar, but NFT price stratification is not solely influenced by rarity. It is perhaps more defined by the community.

For example, mfer and StartCatchers have different Dynamic Traits, and the community defines three dynamic traits >2 >1 >all static. Furthermore, the value point of the mfers project lies in the cultural consensus behind it; the NFT imagery resonates with many holders, evoking a sense of connection.

The influence of rarity on price is greatly diluted, replaced by aesthetic resonance and community consensus. Additionally, NFT prices are also related to celebrity effects; some Trait NFTs associated with celebrities sell for above average prices. For instance, many people choose to buy NFTs with the same Trait as the one held by Jay Chou, thereby driving up the prices of such NFTs.

Thus, it is evident that when issuing NFTs, project teams need to have many ideas. The community's creation of more gameplay means easier promotion and increased circulation. The gameplay (or gimmick) embedded in PFP project images is one aspect that needs to be considered. On the other hand, for those project teams planning to create NFT valuation tools, it is essential to understand that NFT value is determined and influenced by multiple factors. PFP projects redefine rarity while expanding their influence through collaboration and meme dissemination, implicitly increasing NFT prices.

Conclusion

By 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, it is important to note that due to differences in holding time and the timing of buyers' purchases, there are certain biases and variables in the analysis of price and rarity. Additionally, we conducted further analysis based on dynamic factors such as secondary market circulation time, project activities, and differences, helping investors better plan their investment strategies and return expectations in the NFT market.