Mining NFT price manipulation behavior through on-chain data

Introduction

In a recent article, we analyzed those wash behaviors in order to obtain trading rewards. In it, we briefly describe another type of wash trading designed to mislead the market, manipulate market prices, and/or artificially increase trading volume. In this article, we will take DenDekaDan Genesis Omikuji as an example to share our research on NFT market manipulation. In this series, we uncover a series of irregular behaviors that seem to be behind organized and artificially inflated prices.

Money laundering transactions for transaction rewards are usually easy to identify because they are usually back and forth between several fixed wallets. This type of washtrader basically has no incentive to hide the fact that a token is being traded between the same person/entity, mainly because NFT marketplaces (mainly X2Y2 and LooksRare) do not penalize this behavior when handing out trade rewards .

On the other hand, money laundering transactions to create false information about the true value of NFT collectibles are usually carried out in a more covert but organized manner. Parties who may be interested in this type of market manipulation include: NFT project parties, market makers, or whales with sufficient capital and technical expertise to influence the entire market. More importantly, in order to hide the fact that only a few entities may be generating volume and driving up prices through transactions back and forth between a few entities, this type of market manipulation often involves the use of a large number of wallets, which makes only It is difficult to find signs of manipulation behind it when paying attention to the behavior of a small number of addresses.

Methodology

To effectively identify organized market manipulation behind NFT transactions, we focus on analyzing two types of data:

• Transaction Data: This includes on-chain data, such as transactions and prices (transaction prices and floor prices), and off-chain data, such as asks and bids data.
• Address Correlation: This involves researching the correlation of funds between traders, such as historical fund transfers between wallets, whether wallets have the same source of funds, and whether wallets have the same where funds have gone.

By studying transaction data, we can spot suspicious trends in price and volume. By gaining insight into the connections between wallets, we can investigate whether an entity controls a large number of wallets for transactions and examine their behavior.

Case Study: DenDekaDan Genesis Omikuji

We wanted to understand why the floor price jumped from 0.05 E to 2.5 E (a 5000% increase!) after the first week of the series' launch on December 31, 2022. After studying the data, we found that:

• Combining the distribution of floor price trends, transaction prices, and pending order prices reveals a suspicious pattern that may suggest price manipulation behind it.
• Among the addresses that traded this series in the first week, it can be found that many addresses have financial connections.

Finding 1: The floor price shows a regular upward trend

The series was released on December 31, 2022. In the first week after the release, the floor price showed a regular upward trend.

First, around January 1st, the floor price increased to about 1E. For the next four days, from January 2nd to January 5th, the price increases by about 0.5 E every other day (January 3rd, January 5th). After the floor price reached about 2E, it remained in the range of 2-2.5E. This regular uptrend could be a coincidence, but it's also highly suspicious, prompting us to dig deeper.

Figure 1: Floor prices rose regularly in the first week after launch

Finding 2: The distribution of transaction volume and transaction price is also suspicious

Comparing the daily trading volume with the floor price trend, we found that the trading volume on the day when the local board price changed significantly (December 31, January 1, January 3, January 5) was also higher than the local floor price. On days when the price is relatively flat (ie January 2nd, January 4th, January 6th, etc.) the volume is much higher. This seems to suggest that someone behind the scene is executing a large number of transactions to affect the floor price on January 1st, January 3rd and January 5th.

Figure 2: Trading volume soars on a day when the floor price rises sharply.

To test this hypothesis, we further examine the price distribution of all transactions that occurred during the first week. As shown in the figure below (different colors represent transactions on different days), almost all daily transactions occur below the floor price of the day. This discovery is very interesting, because the most direct way to raise the floor price to the target position is to sweep away all pending orders below the target price. This is another sign that there is organized price manipulation behind it.

Figure 3: Distribution of transaction prices in the first week after launch

Finding 3: The distribution of pending order prices is also similar to the distribution of transaction prices

Another important aspect when raising the floor price is to create pending orders (because floor price = lowest asks price). Looking at the distribution of pending order prices, we see that it follows a similar pattern to transaction prices. Specifically, pending order prices move regularly from lower prices to higher prices. Whereas under normal circumstances, when collectibles are first launched, we should see sellers placing orders at a more spread out asking price as there is no consensus on the value of the collection.

Figure 4: Price distribution of pending orders in the first week after launch

Finding 4: Address Correlation

We found that 141 addresses (15%) of all traders in the first week could be linked based on ETH transfer relationships between wallets. Together, these addresses accounted for about 40% of the transaction volume in the first week. It is possible that they are controlled by the same entity behind them, and given the high volume of transactions generated by these addresses, they may be the party behind the price manipulation. (Note: The research in this paper was conducted before we developed an address clustering algorithm that could more accurately identify connections between addresses, so there may be some bias in the address correlation data here. Address clusters are very helpful when investigating potential market manipulation.)

in conclusion

In this study, we share a framework built on our experience and industry knowledge to analyze NFT price manipulation. Although it can reveal some suspicious phenomena and behaviors, we also know that this framework is not complete, and we are always improving our research methods. Since there are few similar studies, we hope to contribute to the NFT field as well as the on-chain data analysis community by sharing our approach. At the same time, we hope that this study will be able to shed light, draw more attention to NFT market manipulation, and promote more research and analysis. We believe that with the joint efforts of the whole community, we can further understand and solve these problems, thus contributing to the sustainable development of this field.

About the author

This analytical study was co-authored by Helena L. and Lin S. of Eocene Research. Follow us on Twitter to stay updated with more of our NFT analysis and research.

Official Twitter account: