Research Summary: NFT Wash Trading: Quantifying Suspicious Behaviour in NFT Markets

[Idara_Effiong]

Great summary @f13r. I enjoyed reading through your work.

In February 2022, [Chainalysis] (NFT Money Laundering and Wash Trading) published a report on NFT wash trading by analyzing sales of NFTs to addresses that were self-financed, or “funded either by the selling address or by the address that initially funded the selling address.” It concluded that “some NFT sellers have conducted hundreds of wash trades.”

In particular, it was able to identify 262 users who had sold an NFT to a self-financed wallet more than 25 times. Although Chainalysis admitted that it could not be 100% sure that all of the users were NFT wash traders, 110 of those wallets collectively generated $8.9 million in profit in 2021.

Furthermore to being safe, there are signs to look out for when an NFT has been wash traded, such as:

1. A major difference between the floor price and list price of an NFT in a collection.
2. Ultimately, researching an NFT collection and
3. Looking at the transaction history is important to understanding if an NFT’s price has been manipulated

[circusqp]

NFT trading combines both art and money markets, both of which are complex on their own not to mention the crypto aspect. The result is a rather complex asset class.

Ironically, NFT markets attract a large segment of population that are not specialized in navigating either money or art markets and might not be in a position to spend enough time identifying adversarial patterns.

This work is a nice resource towards exposing the risks in the market to the growing population of retailers and hopefully make participation more informed and safer.

[WaterLily]

In my opinion,Someone can make this project successful an infinite number of accounts and conduct business between them. The standards of need and proportionality are needed to ensure that this does not violate any privacy or privacy laws. But that might alter when investigators emphasize and also use existing pro laws against fresh NFT platforms. Perhaps NFT marketplaces could intervene in this situation by keeping an eye on questionable transactions and restricting them, or, in the most extreme case, blocking accounts.

[Idara_Effiong]

Thank you @f13r for this fantastic summary. It is indeed an excellent summary because it speaks directly to the quantitative suspicious behavior in NFT markets.

There is no doubt that NFTs are selling like hotcakes (digitally). Christie’s sold a digital collage by artist Beeple for $69.3 million at the NFT auction in March 2021. When you start translating the real into hyper-real, you see endless scenarios that capture people’s imagination. Nevertheless, as with any shiny new toy, there have been fraudsters appearing within the binary code.

The manipulation of NFTs is similar to what we’ve seen with more traditional crypto assets, such as Bitcoins. Hacking wallets and stealing the NFTs is not just another scam, but a process through which it’s possible to assign multiple ownerships to digital designs using the NFTs.

Users have to submit their social media handles for NFT verifications, but the entity submitting these details need not prove ownership. Digital works have been fraudulently mis-sold due to this practice. Did we just create yet another avenue for crypto-crime instead of simplifying it?

Wash trading is pretty widespread on centralized cryptocurrency exchanges, as well as the NFT industry. Our mission is to warn you about the possibilities that keep NFTs a bit too clean. Knowing how to recognize frauds and scams will make it easier to avoid them. Isn’t it? But first, what are NFTs?

How is Wash Trading identified?

Early detection is primarily achieved through the use of graph clustering and nearest neighbor algorithms. But a related phenomenon called collusive cliques has also been researched.

On aggregated order volume time series, hidden Markov models, spectral clustering, and correlation statistics are other methods that help. Check out more about how these models can help in identifying complex sequence data, fraudulent activities, and failure detection.

One of the first studies to explicitly examine wash trades in the markets was by Cao et al. According to them, prior studies focused primarily on collusive and correlated trading patterns rather than wash trading’s more specific form.

A given set of trades consists of subsets that result in no position change to the traders involved. Based on their topological structures, these trades comprise a closed cycle. A closed-cycle trade is one that does not engage in exchanges with other economies. Closed economies are completely self-sufficient, which means no goods are imported or exported.

Although this sounds very much sorted on paper, in practice, it needs an ecosystem that can detect wash trading in NFTs for real. Do we have a solution for that? Yes! And the good news is Scour!

Detecting wash trading the Scour way!

With all the good news of Scour, they have come with incredible approaches to identify the fraudulent activities more precisely and practically. Having said that, with multiple methods on the table, when caught, washtraders keep coming up with new ways to washtrade. New wallets are created and new methods of washtrading are tried.

However, Scour reduced the washtraders by over 40% within two months. Unbelievable? How does it happen?

How does bitsCrunch’s knowledge graph provide a solution to Wash Trading?

As far as we know now, market manipulators are hard to catch. Even though the tokens are non-fungible, they are still anonymous, making fraud detection difficult. But, Scour upgrades every week to identify the new possibilities of wash trading by extreme research on the NFT activities and keeps a checkmate to them asap.

Scour is a unique product created by BitsCrunch, an AI-powered analytics company that can actively remove wash traders from exchanges.

The Scour index compiles transactions, wallet addresses, and distributions of reward tokens. Using knowledge graphs and AI technology, Scour can identify bogus orders and complex patterns of wash trades. The technology not only secures investors and markets but also solves the open secret of cryptocurrencies.

BitsCrunch built Scour with a detailed knowledge graph that includes a complete index of the NFT markets transactions thanks to partnerships with industry leaders like Covalent, Rarible, and deeplearning.ai.

At its core, Scour is an AI product that flags the spoofing activity among the traders to protect the platform (marketplace) from its market-mining operations

[WaterLily]

We identify what we believe may serve as a lower bound estimation for suspicioustrading behaviour on NFT markets, following the definition of wash trading: setsof trades between collusive addresses, without taking market risk, that lead tono change in the individual position of the participating addresses. Our findingsindicate that (I) adversarial agents exhibit a clear preference towards fast andsimple cyclical patterns, (II) the level of suspicious activity varies significantlyacross NFT collections, (III) illicit activity could still be done in a manual fash-ion, and (IV) the activities do not necessarily produce the intended price impact,as other exogenous factors such as age and sentiment are more relevant to pricediscovery.As a theoretical contribution, we add descriptive knowledge to an emergingfield of research where scientific studies are scarce. We contribute to the grow-ing literature on the identification of illicit market behavior in centralized anddecentralized crypto markets, by conducting the first in-depth examination ofNFT wash trading on the Ethereum blockchain. We contribute empirical statis-tics of fraudulent behaviour and a set of suspicious transaction graphs to fosterthe understanding of wash trading in increasingly financialized NFT markets.The valuable insights we generate for practitioners are twofold: First, we providevaluable insights to prevent collectors from buying NFTs that are potentially in-flated by wash trading. Second, we discuss practical countermeasures increasingthe standards for the wider NFT ecosystem. Further research opportunities are manifold and include studying NFT markets on other blockchains as well asresearching the correlation between suspicious behaviour and sentiment data.

[Freakytainment]

@WaterLily

Our findingsindicate that (I) adversarial agents exhibit a clear preference towards fast andsimple cyclical patterns, (II) the level of suspicious activity varies significantlyacross NFT collections, (III) illicit activity could still be done in a manual fash-ion, and (IV) the activities do not necessarily produce the intended price impact,as other exogenous factors such as age and sentiment are more relevant to pricediscovery.

Nice write up, Enjoyed reading. But should have been better if you had arranged this properly for easy understanding. Seems like a rush work.

[Amazingdez]

Nice summary @f13r I really enjoyed reading it… a little contribution and a question though!.

To confirm ownership and validity, non-fungible tokens (NFTs), a cryptographically secure, unique digital identification that is maintained in a blockchain, are employed. Additionally, they can represent tangible goods like artwork and real estate by being “tokenized,” which increases trading efficiency while reducing the danger of fraud. To an extent, NFTs most popular use are Digital collectibles as they enable non-duplicable in-game assets, True…

My question is: what is the possibility of NFTs non-duplicability(not being able to Copy) surviving in the nxt 5years, because of the thriving rate of technological development??.

[f13r]

Thanks @Amazingdez!

So TECHNICALLY the underlying cryptographic security of blockchains like Ethereum is unlikely to be threatened by new developments in quantum computing within the next 5 years. Should there be any known threat, Ethereum will most likely undergo a hardfork to change the underlying cryptography to safer methods.

However, from a SOCIAL point of view the picture is less clear - the same NFT can theoretically exist on many different blockchain networks. Efforts have been taken to create global NFT identifiers that work across blockchain networks. The question remains if people will adopt such proposed standards…

I hope this answers your question?

[HomicidalChicken]

Great summary, surprising to see the volume of measurable wash trading is lower than I expected. Hermes_Corp raised some excellent points on the potential methodological pitfalls of the author’s approach, I agree that there is more to be learned on the practice through better knowledge of wash trading patterns in NFT markets.

One question I’m left with is how much overall impact these wash trades have on the market, beyond the individual transaction. For example, in collections that have received more wash trading, does the price effect extend to other NFTs that were not wash traded?

In other words, because of high psychological association between NFTs in a collection, is the effect more similar to “painting the tape” than wash trading alone? In that case, how does the effect impact the overall profit extraction of an identified wash trader in the dataset? Others have pointed out floor price manipulation, which seems like a highly relevant factor when considering the overall prevalence and impact of price manipulation tactics on the market.

I think it would be interesting to see this data categorized in terms of profit extraction (i.e closed loop trades ending with an external sale) and not just trading volume, in order to see how these trades distort the market. Given the large degree of inequality between the profitability of NFT traders, even a low relative percentage of manipulated trades may be responsible for an outsized degree of extraction from market participants.

[Amazingdez]

Yes it does, from a soicial point of view…thank you @f13r

[DIVINE]

Hello @f13r nice job, just wanted to briefly comment on how to detect NFT wash Trading which based on my understanding of this paper and as well my practical experience . I have looked at how a few analytics platforms, including Footprint Analytics, do their detection and followed their logic. Their methodologies are somewhat similar, to be honest. Along with my own knowledge and analysis, here is a checklist of suspicious data and activity that should trigger any prospective NFT buyer’s alarm bells:

1. A particular NFT is traded by the same address more than X times a day while the rest of the collection remains untouched.

2. The same address is trading the same NFT in a high-frequency manner.

3. A collection of NFT goes into a self-selling in a high-frequency manner when there is no marketing or promotion backing the sale.

4. The average historical price transacted is X times higher on marketplace A vs. B.

5. The sale price of an NFT is transacted X times higher than the lowest-priced NFT available for sale.

6. The same wallet addresses funding all the suspicious wallets that buy and sell the NFTs.

7. An abnormal high trading volume on a constant basis.

The above assumptions are not perfect, and I hope to work with researchers on developing a more comprehensive scorecard to determine NFT trends and behaviors more effectively. The ability to trace multiple wallets over time to identify various levels of relationships would be vital too.

[Idara_Effiong]

@Henry to the second part of your comment, It is possible that wash trading could have affected the NFT (non-fungible token) market to some extent. NFTs, which are digital assets that are unique and cannot be exchanged for other assets on a one-for-one basis, have gained popularity in recent years as a way to represent and trade digital assets such as art, collectibles, and other items of value. However, the NFT market is still relatively new and immature, and there may be instances of wash trading or other forms of market manipulation occurring within it.

It is important for market participants to be aware of the potential risks of wash trading and other forms of market manipulation, and to take steps to protect themselves and their investments. This can include carefully evaluating the market and the assets in which they are considering investing, as well as using trusted and reliable sources of information and market data.

[bparlan]

I use Covalent onchain data analysis tool, will utilize these methods. I’m not into nft’s, but can access like 40 chains together, so if you got any NFT suggestions, I would gladly check them to see results of those 7 steps. Personally I love Keepers project, got KPR #425 for full disclosure, I believe I will check those on that definitely! (: Thanks for roadmap!