Using Machine Learning to Detect Cryptocurrency Scams

Callum Haine


Supervised by Amir Javed; Moderated by Neetesh Saxena

The last year has seen an exponential increase in the value of many popular cryptocurrencies such as Bitcoin and Ethereum. This has inevitably led to increased media attention, and a flood of new retail investors wanting to grow their portfolios in this lucrative new market. This, coupled with a current lack of regulations, has unfortunately left the market rife with scams. These usually take the form of small-cap cryptos being promoted heavily with the intention of drawing in investment; artificially increasing the price of the currency. The scammers then sell their pre-bought crypto at the inflated price, crashing the value and leaving those caught in the scam out-of-pocket. The aim of this project is to investigate and implement a machine-learning model capable of monitoring Twitter data, and flagging any potential scams. The model should also take into account current market cap and daily volume of crypto tokens in order to make better classifications. This model can then be used to provide a live dashboard, which displays data on potential scams.

Final Report (05/11/2021) [Zip Archive]

Publication Form