Researchers from Imperial College London have developed an algorithm which they say can be used to predict cryptocurrency “pump and dump” schemes, something that allegedly contributes to USD 7 million worth of trading every month.
Pump and dump is a form of insider trading that has become increasingly commonplace since the rise of tokens in the cryptocurrency industry. It begins with an organizer accumulating a large amount of an obscure token in private, before announcing it to a closed group which then begin purchasing it also in large amounts, spiking the value.
Within minutes the selloff begins, with participants making an easy profit off those slow to the game and unaware of what was really going on.
While pump and dump schemes are present in conventional commodities trading, it is tightly regulated and participants can find themselves is significant trouble with the law. The issue in the cryptocurrency market is that there is no central authority to crack down on those responsible, hence these types of activity continue to take place, usually organized on private channels such as Telegram.
Jiahua Xu and Benjamin Livshits from Imperial College London have used machine learning to give investors a way to spot the schemes themselves, with some mathematical legwork.
Looking at 236 pump and dump cases, the researchers noted that the targeted cryptocurrency was nearly always preceded by unusual buying activity as the organizer accumulated a store prior to the pump. “The study reveals that pump-and-dump organizers can easily use their insider information to take extra gain at the sacrifice of fellow pumpers,” they explained.
These schemes can be avoided by identifying unusual buying patterns in obscure coins, Xu and Livshits identified, backing up their theory by training a machine learning algorithm with historical pump and dump data.
This is not a final solution, however, as organizers of these scams may change the patterns of their behavior to avoid detection by the algorithm.
Follow BitcoinNews.com on Twitter: @bitcoinnewscom
Telegram Alerts from BitcoinNews.com: https://t.me/bconews
Want to advertise or get published on BitcoinNews.com? – View our Media Kit PDF here.
Image Courtesy: Pixabay
