Art market information can be mined from historical data on trades of an artwork, if available. We assume a general setting in which artworks are created by artists while collectors can directly buy the artwork at the set price or make an offer (a bid) for it. Sold artworks remain tradable on the secondary market. The focus of our rating system is on artworks; different units like artists, collectors, and even entire galleries are collections of artworks and can be assessed in terms of the artworks they contain. In this setting, there are two major signals of market success for an artwork:
We used the following metrics to assess the bid and sale history of an artwork:
For example, suppose the bid and sale history of an artwork is as follows:
Here, collector A is willing to pay 6 (their largest bid) for the artwork, collector B would pay 4, and collector C would spend 5. Hence, the sum of largest bids on the artwork is \(\beta = 6 + 4 + 5 = 15\) and represents a sort of open interest for the piece. The total number of bids is \(\gamma = 6\) and the number of different bidders is \(\delta = 3\) (A, B and C). The sum of all sales is \(\alpha = 5 + 6 = 11\).
We thus define an artwork rating \(\rho\) for an artwork \(t\) as a weighted average of the above four metrics after normalization:
\[ \rho(t) = \frac{1}{3} \cdot \frac{\alpha(t)}{\max(\alpha)} + \frac{1}{3} \cdot \frac{\beta(t)}{\max(\beta)} + \frac{1}{6} \cdot \frac{\gamma(t)}{\max(\gamma)} + \frac{1}{6} \cdot \frac{\delta(t)}{\max(\delta)} \]
Given a collection of artworks \(S\), we further define the cumulative rating \(\sigma\) of \(S\) as:
\[\sigma = \sum_{t \in S} \rho(t)\] and the average rating \(\mu\) of \(S\) as:
\[\mu = \frac{\sigma}{|S|}\] where \(|S|\) is the number of elements of \(S\). Notice that \(\sigma\) depends on the size of the collection \(S\) while \(\mu\) is size-independant.
An important caveat is how to assess the actual price of sales and bids. Since digital artworks are mainly traded in crypto currencies (mainly Ether, the coin of Ethereum blockchain), and these coins are not stable (they show large variance of the historical prices), we decided to use the price expressed in fiat money (dollars) at the exchange rate of the time of the bid or sale.
We applied our method to the entire collection of SuperRare crypto art gallery (data from 5th April 2018 to 2nd November 2020). We first computed the artwork rating \(\rho\) for all artworks of the collection.
rank | name | artist | rating | sale volume | bid volume | bids | bidders | id | link |
---|---|---|---|---|---|---|---|---|---|
1 | Rebirth of Venus | NA | 0.851 | 88085 | 305085 | 35 | 21 | 16297 | click |
2 | AI Generated Nude Portrait #1 | videodrome | 0.592 | 123544 | 128735 | 13 | 10 | 1 | click |
3 | The Innovator’s Dinner | fewocious | 0.521 | 0 | 175847 | 39 | 26 | 13670 | click |
4 | AI Generated Nude Portrait #5 | videodrome | 0.440 | 122620 | 14663 | 10 | 8 | 65 | click |
5 | AI Generated Nude Portrait #3 | videodrome | 0.430 | 101898 | 103053 | 4 | 4 | 3 | click |
6 | Hurt Feelings | fewocious | 0.340 | 8834 | 54200 | 31 | 20 | 13907 | click |
7 | Latent Space of Landscape Paintings #1 | videodrome | 0.322 | 798 | 130129 | 21 | 14 | 135 | click |
8 | Elephant Dreams | rac | 0.312 | 25962 | 68554 | 20 | 13 | 14316 | click |
9 | Möbius Knot | pak | 0.307 | 44352 | 102169 | 12 | 4 | 16428 | click |
10 | shutdown –reboot | NA | 0.281 | 24529 | 68196 | 23 | 7 | 16647 | click |
11 | AI Generated Nude Portrait #7 Frame #175 | videodrome | 0.279 | 4929 | 3361 | 32 | 20 | 365 | click |
12 | CYBERSNEAKER | rtfktstudios | 0.271 | 11442 | 53936 | 28 | 10 | 14504 | click |
13 | Dharma Dragon | android_jones | 0.234 | 21425 | 45401 | 15 | 10 | 14834 | click |
14 | High Functioning | killeracid | 0.228 | 0 | 6129 | 27 | 17 | 47 | click |
15 | AI Generated Nude Portrait #7 Frame #153 | videodrome | 0.228 | 2155 | 7780 | 39 | 8 | 343 | click |
16 | “One of Us” Variation 1 | mattkane | 0.220 | 1464 | 3984 | 40 | 7 | 5051 | click |
17 | UAP - Unidentified Art Phenomenon | coldie | 0.217 | 19138 | 50647 | 14 | 8 | 9343 | click |
18 | DystoPunk 3D | coldie | 0.216 | 6505 | 33712 | 28 | 7 | 7168 | click |
19 | The Frame | pak | 0.216 | 14577 | 57641 | 18 | 6 | 11958 | click |
20 | Roarrr!! | suryanto | 0.215 | 5907 | 25062 | 29 | 8 | 16768 | click |
Then we assessed artists by the artworks they created. Here, we have two choices. We can assess an artist using the cumulative rating of all artworks tokenized by the artist. This choice, however, favors the most productive artists. Since tokenization is (almost) free when you are a white-listed artist on a gallery, we do not make this choice. The second possibility is to assess an artist using the mean of the ratings of all artworks they created. However, different artists create at different rates (there are artists that tokenize a new piece each day and others that mint one new artwork every month) and, moreover, they have different histories (some have long been active in the space while others just landed there). It turns out that the collections of artworks created by the artists are very heterogeneous in size. It is not statistically sound to compare means over samples of sizes that differ largely. Hence, we adopted a top-n-min-k approach. Given numbers \(k\) and \(n\) with \(k \geq n \geq 1\):
A high value for \(n\) favors artists with a long activity history; on the other hand, a small value for \(n\) is inclusive with respect to artists with a short activity history, including emerging ones. We set \(n = k = 20\).
rank | artist | rating | SuperRare | OpenSea |
---|---|---|---|---|
1 | videodrome | 0.16023 | click | click |
2 | pak | 0.15819 | click | click |
3 | hackatao | 0.12173 | click | click |
4 | coldie | 0.10124 | click | click |
5 | xcopy | 0.10065 | click | click |
6 | twistedvacancyart | 0.07987 | click | click |
7 | glasscrane | 0.07905 | click | click |
8 | osinachi | 0.07820 | click | click |
9 | alotta_money | 0.07669 | click | click |
10 | jenisu | 0.07319 | click | click |
11 | frenetikvoid | 0.07235 | click | click |
12 | osiris | 0.07175 | click | click |
13 | gric | 0.07008 | click | click |
14 | missalsimpson | 0.06871 | click | click |
15 | carlosmarcialt | 0.06592 | click | click |
16 | mattkane | 0.06502 | click | click |
17 | suryanto | 0.06495 | click | click |
18 | goldweard | 0.06347 | click | click |
19 | sveneberwein | 0.06344 | click | click |
20 | _totemical | 0.06192 | click | click |
We finally assessed collectors with the same method, considering the collection of artworks they acquired and setting \(n = k = 50\).
rank | collector | rating | SuperRare | OpenSea |
---|---|---|---|---|
1 | thevault | 0.08081 | click | click |
2 | moca | 0.07945 | click | click |
3 | ethsquiat | 0.07201 | click | click |
4 | moderats | 0.07124 | click | click |
5 | maxstealth | 0.06683 | click | click |
6 | basileus | 0.06555 | click | click |
7 | gltr | 0.05753 | click | click |
8 | tokenangels | 0.05520 | click | click |
9 | randaartvault | 0.05332 | click | click |
10 | 0x123456789 | 0.05287 | click | click |
11 | coldie | 0.05140 | click | click |
12 | matrix | 0.05022 | click | click |
13 | blockchainbrett | 0.05011 | click | click |
14 | momuscollection_sr | 0.04886 | click | click |
15 | jedscryogenicstorage | 0.04666 | click | click |
16 | bitbuzz | 0.04553 | click | click |
17 | deej | 0.04524 | click | click |
18 | mantaxrartvault | 0.04320 | click | click |
19 | NA | 0.04293 | click | click |
20 | zonted | 0.04043 | click | click |