To the moon: defining and detecting cryptocurrency pump-and-dumps

[Audio] Good morning everyone. Today, I would like to present to you our yearly analysis on cryptocurrency pump-and-dump schemes. We will be discussing techniques for detecting and preventing these schemes and provide you with criteria to inform future decisions. Let's get started..

[Audio] A pump-and-dump scheme is a type of fraud where the perpetrators build up a commodity and then artificially inflate the price through spreading false information before selling what they bought at a higher price. Cryptocurrencies, such as Bitcoin, are digital mediums of exchange that rely on cryptography instead of a central institution. This PowerPoint slide explains the concept of cryptocurrency pump-and-dump schemes and the background behind them. It proposes criteria for identifying cryptocurrency pump-and-dump schemes and an automated anomaly detection approach to locate suspicious transactions. Furthermore, it focuses on the exchange and cryptocurrency pairings level, intended to spark an academic interest in the topic..

[Audio] Looking at the phenomenon of Pump and Dump schemes in the cryptocurrency context, a Pump and Dump Scheme occurs when an individual or a group artificially boosts the price of a certain commodity, in this case a cryptocurrency, in order to later sell it off to mislead buyers. This is an example of stock market manipulation and has been illegal in the United States since 2002. The Pump and Dump has evolved in the cryptocurrency context, with dedicated public groups forming within online chat rooms like Discord and Telegram, solely for the purpose of organising pump and dump schemes on select cryptocurrencies. Reports indicate that prices increases of up to 950% have been witnessed, demonstrating the extent of manipulation these groups are capable of. For these groups to be most effective, they usually target less popular coins with low market cap and circulation, since these are deemed easier to manipulate. Crime prevention strategies should be informed by the criteria for a cryptocurrency Pump and Dump Scheme..

[Audio] Cryptocurrency pump-and-dump schemes are gaining attention because of their potential to cause harm. To identify their patterns, it is necessary to develop a definition that takes into account the characteristics they share with traditional penny stock pump-and-dump schemes, such as the targeted assets, tactics, and timescales. With this definition in hand, we can work to prevent and address the issues caused by these schemes..

[Audio] Data provided by cryptocurrency exchanges can come in the form of Open-High-Low-Close Volume (OHLCV) entries. These create a chart representation of the trading data for a specific time period which can be used to detect potential pump-and-dump patterns. OHLCV allows one to examine the price and volume of a specific symbol pair (such as BTC/USD) over a given period and detect any increase in activity that could indicate manipulation. Market cap, the number of exchanges, and the currency the coin is trading with can be considered when evaluating a potential pump-and-dump. All of the needed data can be obtained through the CCXT library, making it available to anyone..

[Audio] We discussed the automated detection of potential cryptocurrency pump-and-dump schemes using an unsupervised approach to anomaly detection which identifies non-conforming points in a dataset. This can correspond to price and volume spikes, indicative of a pump-and-dump. Kamps and Kleinberg used a filtering process to identify 24 potential exchanges, while filtering further to identify exchanges with at least 50 symbol pairs and at least 20 days of historical 1-hour OHLCV data. This enabled them to identify five exchanges with 480 candles to analyze for indications of a pump-and-dump, as well as the three major types of anomalies; point anomalies, collective anomalies and contextual anomalies..

[Audio] Anomaly detection is an important part of understanding cryptocurrency pump-and-dump schemes. It is broken down into two components: price and volume anomalies. A price anomaly is when the high price at any given point is greater than the computed anomaly threshold for that point; volume anomaly is defined similarly, but with the moving average computed using the volume of the cryptocurrency. To detect pump-and-dump schemes, it is necessary to detect the co-occurrence of both a price anomaly and a volume anomaly. Contextual information such as the market capitalization of the cryptocurrency can also be used to better detect pump-and-dump schemes. To determine the optimal parameters, different values must be tested and the results used to suggest balanced parameters that can detect pump-and-dump schemes. With the right parameters, potential pump-and-dump schemes can be accurately located in cryptocurrency..

[Audio] Anomaly detection was used to detect potential pump-and-dumps. Three parameters were tested - initial, strict and balanced. Initial parameters had an estimation window of 12 hours, a 25% volume increase and a 3% price increase. The strict parameters used an estimation window of 24 hours, a 400% volume increase and 10% price increase. Balanced parameters had an estimation window of 12 hours, a 300% volume increase and a 5% price increase. Results showed the balanced parameters to be the most effective, detecting 1.6 pump-and-dumps per symbol and 75% of alleged pumps followed by a corresponding price dip. These parameters may be successful in indicating a pump-and-dump scheme..

. . Page 9 of 18 Kamps and Kleinberg Crime Sci (2018) 7:18.

[Audio] We examine two cases where our system was able to successfully detect potential pump and dump schemes and two cases where it could not. In the successful detection cases, we observe a corresponding rise in both price and volume at the announcement time of the pump, which our system was able to flag. In the two unsuccessful detection cases, there is also a rise in both price and volume, yet our system was not able to mark them as fraudulent. This analysis highlights the importance of the right settings when it comes to anomaly detection..

. . Page 11 of 18 Kamps and Kleinberg Crime Sci (2018) 7:18.

[Audio] This paper examined techniques from anomaly detection to detect potential crypto currency pump and dumps schemes and proposed criteria to inform future crime prevention. Breakout indicators and reinforcers were identified as criteria to locate pump and dumps, with data investigated using an anomaly detection approach. Results found that the top 10 symbols with the most pump and dumps were mostly located on the Lbank and Kucoin exchanges. Additionally, pump and dumps were sometimes observed days after one another, likely attributed to the pump and dump group causing a momentum. To correctly identify these cases, in which the price maintains momentum for some time after the announcement, the proposed algorithm can be improved to include decreasing volume, in addition to price dips..

[Audio] Kamps and Kleinberg (2018) combined anomaly detection techniques with criteria for a cryptocurrency pump-and-dump, and identified evidence of clustering in the data. Analysis revealed that a small portion of coins were strategically targeted more frequently than others, likely due to a repeat victimisation phenomenon. To prevent these schemes, strategies to increase effort and risk such as additional verification for vulnerable coin pairs, and automated detection of anomalous trading with human observers, are recommended..

[Audio] The findings of Kamps and Kleinberg suggest that prevention of cryptocurrency pump-and-dump schemes requires a collaborative effort between private organizations, such as cryptocurrency exchanges, and government organizations. It is further noted that an increase in government regulation may present a challenge as it could undermine the fundamental idea of a decentralised and government-interference free cryptocurrency trading. Furthermore, it is acknowledged that an interdisciplinary problem-oriented approach is key to mitigating these pump-and-dump scheme, and future research is needed due to the limited number of days in the dataset.The findings of Kamps and Kleinberg suggest that prevention of cryptocurrency pump-and-dump schemes requires a collaborative effort from both the private sector, such as cryptocurrency exchanges, and government entities. Additionally, more research is needed due to the limited number of days available in the dataset which is necessary to further understand the issue of pump-and-dump schemes. Moreover, an interdisciplinary and problem-oriented approach is seen as a potential solution for mitigating this issue. Finally, government regulation may present a challenge as it could potentially undermine the fundamental idea of a decentralized and government-interference free cryptocurrency trading..

[Audio] Kamps and Kleinberg (2018) suggest that detecting potential cryptocurrency pump and dump schemes is a difficult task that requires a balance between false positives and false negatives. To this end, having the right parameters for the detection algorithm is essential for successful detection. Building a database of confirmed pumps for testing the accuracy of the detection algorithm could prove to be an interesting research direction. As demonstrated in Figure 10, a successful detection of a pump and dump scheme results in anomalous price and volume spikes that are highly visible..

[Audio] Using anomaly detection as a tool can help to identify possible cryptocurrency pump-and-dump schemes. This research has proposed criteria for such a scheme, however, further research needs to be done to understand how these schemes are organized and implemented, as well as how to prevent them. It is important to investigate these issues in a timely manner..

[Audio] Kamps and Kleinberg's study in Crime Sci for 2018 outlined techniques for detecting potential cryptocurrency pump-and-dump schemes, and proposed criteria for detecting pump-and-dump incidents. The authors collected data from five different exchanges, covering 302 symbol pairs and 480 hours of 1 hour candles. A chart, depicting the results of a pump-and-dump promoted by the group Crypto Trading™, is given in Figure 12 and an overview of the obtained data is given in Table 7. The data and code needed to reproduce the findings can be found at the provided link..

[Audio] I have discussed the research of Kamps and Kleinberg, which examines techniques from anomaly detection to locate potential cryptocurrency pump-and-dump schemes, and proposes criteria for a cryptocurrency pump-and-dump to inform future crime prevention. This research has provided invaluable insight into the problem of cryptocurrency pump-and-dump schemes, and I hope it can be used to protect investors from scams. Thank you for listening..