The continuous debate revolves around determining the genuine market value of NFT tokens, particularly in the realm of generative art. This process is complex due to the myriad of variables at play, including the state and activity of the market, the collection’s and artist’s popularity, floor price fluctuations, premium traits, rarity ranks, outliers, and the market demand for specific traits.
The current reliance on the floor price as one of the primary methods to evaluate most of the tokens is flawed, primarily because it overlooks several critical factors, notably the inherent value of specific and premium traits, liquidity concerns, and the potential for floor price manipulations. With X, Y, and Z in mind, what is the fair market price of a particular token?
That’s why we are thrilled to unveil A-valuation, an engine designed to transform decision-making and portfolio management practices. “A” means “Artacle”, though you are free to embrace its alter ego, “Alpha”.
Ground truth
Machine Learning typically performs best when provided with a clear ground truth value for prediction. However in our case, estimating NFT token values is challenging due to incomplete sales data, subjective pricing, and a lack of control over trait values (compounded by non-using of suitable tools like Trait-to-Floor Multiple on Artacle). Moreover, sellers’ biases and underpricing/overpricing add to the complexity of the process.
Through the identification of fundamental distribution laws, we successfully generated a reliable ground truth and developed a set of models for the A-valuation. It is important to bear in mind that the A-valuation must not always align with sales or listing prices, given their subjective nature. And remember that you should also explore the neighborhood of the estimated value by doing your own research with help of the Artacle tools, especially when making important decisions.
To ensure accuracy, the reliability of data is crucial. In our efforts to enhance analysis, we are also integrating additional data sources, including Sotheby’s sales, and we’re incorporating OTC sales data soon.
And of course our favorite rarity ranking plays a very interesting role…
Rarity ranking
While some collectors may overlook rarity ranks, it’s understandable that each person has their own goals, and personal preferences for a work of art matter most. However, when it comes to investing, it’s prudent to consider rarity ranks, as they undeniably play a significant role in pricing.
Indeed, rarity ranks may not be significant for every single collection. It is essential to recognize that suitable tools might not be available always when needed. And when investing was done, reevaluating decisions is not always straightforward. Moreover, various designs of traits can lead to both more and less interesting combinations, resulting in more scarcity in some cases. BTW this is how we calculate rarity ranks.
What we discovered is fascinating. When we consider any subset with a significant number of collections, eliminate outliers and apply necessary normalizations, we consistently observe the same rarity distribution law. This pattern holds true even when combining generative art with PFP collections, resembling the laws of nature. Hence, we cannot overlook rarity ranks in A-valuation of tokens, and fortunately, we have found an effective approach to incorporate them.
A-valuation of tokens
The most prevalent approach to A-valuation:
A-valuation = Top Trait Valuation + R(r, c),
where Top Trait Valuation is based on TTF (a Trait-to-Floor) multiple value of a top trait considering recent market cycle and interdependencies between traits, and R(r, c) is a rarity reward function with the following arguments:
- r — rarity rank
- c — total amount of tokens in the collection
R(r, c) may be negative, in this case it becomes a penalty function. It is important to note that Top Trait Valuation is aligned with common market indicators such as cost basis and protected against price manipulations.
Due to special cases below it is important to note that in certain scenario A-valuation can be substituted.
Special cases
There are tons of special cases where we must resort to individual models. Some examples include scenarios with no traits, no sales history for specific traits, inadequate floor, large distinct count traits and many more.
Another example is the evaluation of SuperRare tokens, which poses its own set of challenges. For this, we have adopted an approach based on artists and the last sale price, particularly for 1/1 artworks. However, we are continuously working on developing better models tailored to this specific context.
Traits valuation
In addition to token A-valuation, it is also feasible to monitor the valuation of individual traits.
Portfolio
Using A-valuation portfolio can be managed more professionally and precisely, reflecting more fair value of tokens in it.
Future work
The following amazing functions are coming soon:
- Desirability ranking. Indeed, we are moving away from rarity ranking and introducing desirability ranking, also known as A-rank. This new ranking will encompass both rarity rank and market value, combining them through A-valuation. BTW it’s already available, read more!
- Best listings. We’ve launched a new dashboard that will display the best listings comparing them with A-valuation. For example, it highlights listings that are approximately 30% lower than their market value, providing valuable insights for potential buyers.
If you have any feedback or new feature request feel free to drop us an e-mail at feedback@artacle.io. Join our Discord and stay tuned!
