1811.10109

[Audio] This paper presents a comprehensive study on pump-and-dump activities in the cryptocurrency market and is authored by Jiahua Xu from École Polytechnique Fédérale de Lausanne, Imperial College London, Harvard University, UCL Centre for Blockchain Technologies, and Brave Software. In it, we discuss the anatomy of a cryptocurrency pump-and-dump scheme, including an analysis of patterns, a predictive model for targeting coins for pump-and-dump operations, and a trading strategy to generate high returns. We now move on to discuss the example of John and Mary..

. . BVBCoin / arc '300 8VB 1—26 o cooze: 2 Official WAf« Pump S. „ The next post be coin to buy Signals lsoo 269* mocK EW-y dTC,mvB 35 Vie of lag of Dumo on Cryptoco site, this a 228* to N at æing wning Nt 300*'.

[Audio] Pumps and dumps on Telegram are fairly easy to execute. A few days before the pump, admins will post the information on the exact time and exchange, as well as the pairing coin. Members of the group should move their funds in the form of the pairing coin into the exchange before the pump time. On the day of the pump, the admins will send countdown messages and the rules of the pump. When the pump begins, the coin's price will usually surge for the first minute, and then the admins will advise members to panic sell. However, there is always the risk that the price of the coin could move in an unanticipated way before the pump, leaving the admin with the advantage of having purchased the coin at a lower price and potentially leaving the other members of the group at a disadvantage..

[Audio] Cryptocurrency has revolutionized the way people buy and sell, enabling speculation. Pump-and-dump activities involve groups of people using hype and misinformation to artificially inflate the prices of certain coins and then reselling them with a profit. In this study, Jiahua Xu provides an analysis of patterns, a predictive model for targeting coins, and a trading strategy to maximize returns. The ethical and legal implications of pump-and-dumps must be taken into account, as well as the lack of clear regulations surrounding them. Case studies can help investors understand how pump-and-dumps work and guard against potential losses. Knowing about these activities is critical to making wise investments..

[Audio] Figure 5 shows the gap between buy and sell volumes during the BVB pump-and-dump. The buy volume surpasses the sell volume, whether based on the target coin BVB or on Bitcoin. This discrepancy between the buy and sell volume signals a greater aggressiveness in buying, likely caused by investors being stuck with worthless coins after the crash. The total number of transactions is surprisingly low compared to the number of channel members, indicating that many investors are either unwilling to engage or aware of the difficulty in securing a profit from a pump-and-dump. From this, we can conclude that the BVB pump-and-dump is a risky investment..

[Audio] We focus on the sources of data for our study. PumpOlymp was used to identify hundreds of pump-and-dump channels, which was then validated using an independent manual search. The official Telegram API was used to retrieve message history from those channels. PumpOlymp’s API was used to gather over 400 pump-and-dump events. Further, CoinMarketCap and CoinGecko provided additional information on features and price movements of coins. The data was then used to analyze and develop a predictive model for pump-and-dump activities..

[Audio] Analysis of pump and dump activities at Binance, Bittrex, Cryptopia and Yobit exchanges reveals some interesting features. Pre-pump volume for coins targeted for pump and dump operations is relatively low, but the volume during the pump is significantly higher than before. Coins used for pump and dump are typically low market cap coins with larger total supply, but comparatively lower ratings than other available coins. These findings could provide useful insight into the pump and dump phenomenon..

[Audio] An analysis has revealed an astonishing level of artificially generated trading volume at 8,793 BTC (93% from Binance) over eight months, amounting to 50 million USD. In particular, it has been observed that the pumps in Yobit and Cryptopia are more effective than those in Bittrex and Binance. It is estimated that admins have generated a net profit of 199.52 BTC, 1.1 million USD, through 348 pump-and-dump activities with an average return of 18%. These results demonstrate that pump-and-dump activities remain profitable and efficient, thus investors must be aware of them..

[Audio] I'm writing to request an extension on the deadline for the project. The current date is April 28th and the deadline is May 3rd. Given the current circumstances, I am not certain I will be able to meet the deadline. I understand the importance of meeting the timeline and would be grateful if I could be granted an extension..

[Audio] The speaker will cover the features of the predictive model for coins on the Cryptopia exchange, which consists of 46 market-related features, such as market capitalization, hourly log returns, and volume. These features allow investors to detect pump-and-dump activities, providing information on which coins have the highest potential for profit. Through this model, a comprehensive overview of pumped coins and the prospect of their future returns can be obtained..

[Audio] We present the specifications of a model used to predict pump-and-dump activities in the cryptocurrency market, testing both classification methods like Random Forest, and logistic regression models like Generalised Linear Models. To ensure proper data handling, a stratified sampling method is adopted, and the LASSO technique is used to pick out the most important variables. The model reveals the coin market cap and the last hour return before the pump are the two main features used to predict which coins will be affected by pump-and-dump activities..

[Audio] The results of the analysis of two models used to assess the accuracy of predicting potential coins to be pumped are demonstrated in this slide. While the RF model performs better on both precision and F1 measure across a wide range of thresholds, the GLM model performs worse, with an area under the ROC curve between 0.63 and 0.88, and thus has been eliminated from further consideration. An investment strategy was created in which coins whose predicted pump likelihood surpass a predetermined threshold are purchased before the actual coin announcement, allowing investors the potential to benefit from this research..

[Audio] Studying and understanding the patterns of pump-and-dump activities is essential to identify and target coins for trading operations. To that end, Jiahua Xu has developed a predictive model for pump-and-dump operations. Validation sample results have shown that the model offers the highest return, up to 80%, when using RF1 with a threshold of 0.3. The test sample results confirm that this model and threshold choice is the most effective in generating a high return, yielding a return of up to 140%. Additionally, market depth has been analyzed to determine the optimal quantity of investments in BTC..

[Audio] This table displays data collected from observing pump and dump events on Telegram over a two and a half month period from June 17, 2018 to February 26, 2019. The table shows the coins which were purchased based on a predicted pump likelihood of the model with a threshold of 0.3, and includes the investment weight, amount of assumed profit, and return on investment. Investment returns amounted to an impressive 60% during the test sample period..

[Audio] A comprehensive study on pump-and-dump activities in the cryptocurrency market has revealed their persisting nature, which drives tens of millions of dollars of phony trading volumes each month. Evidence suggests market movements prior to a pump-and-dump event can contain information which coin will be targeted. Models using machine learning are able to accurately predict pump-and-dump targets, enabling a trading strategy with a potential return of 60%. This study sheds light on the pump-and-dumps within the crypto-market and provides essential information for individuals and regulators to help curb market abuse and criminal behavior..

[Audio] This presentation will explore pump-and-dump schemes in the cryptocurrency markets. We will look at common patterns of pump-and-dump activities and provide a predictive model for targeting coins. A trading strategy to generate high returns by engaging in pump-and-dump activities will also be discussed. Supporting evidence is drawn from sources such as academic papers, regulatory organizations, and online resources. These sources include papers from institutions such as École Polytechnique Fédérale de Lausanne, Imperial College London, Harvard University, UCL Centre for Blockchain Technologies, and Brave Software. To demonstrate the efficacy of our model, the Bitcoin, Ethereum and other cryptocurrency markets are analysed from a variety of perspectives. Government organizations such as CFTC, FTC, and SEC are also consulted. Finally, evidence from journals such as Economics Letters, Journal of Trading, and Harvard Business Law Review, and reports from Business Insider are presented. This evidence is used to demonstrate the importance and relevance of our work on pump-and-dump activities in the cryptocurrency market..

[Audio] Appendix slide presents the results of two statistical models that were used to analyze pump-and-dump activities in the cryptocurrency market. Table and Gini coefficient display the features that were identified as most influential by the random forest model. GLM model provides additional insight, with the intercept coefficients indicating the risks associated with participating in pump-and-dump activities and the significant variables of the model. Understanding the dynamics of pump-and-dump activities facilitated by these results helps investors make better-informed decisions. Thank you for your attention..