Crypto Safety Check: Spotting Risky Ones with AI

Cryptocurrencies have been grabbing headlines and investor attention lately. But before diving in, it's crucial to understand what makes crypto prices tick and which ones might be too risky. This piece explores how AI can help by analyzing historical data and on-chain parameters. You know, those bits of info like total supply and trading volume. The researchers looked back at data from various cryptocurrencies and found some interesting patterns. For instance, max supply and total supply seemed to have a negative effect on prices. On the flip side, higher trading volumes over 24 hours saw a slight boost in prices. But hold on, there's more! They found that a whopping 39% of these digital currencies vanished from the market, while a lucky 10% stuck around for over 1000 days. To make things even more complex, they grouped cryptocurrencies into five different clusters based on their on-chain parameters. This clustering helps investors get a clearer picture of a crypto's character compared to its peers. Finally, they brought in some AI smarts to play guessing games. They used different classifiers to predict if a crypto was risky or not. The winner? K-Nearest Neighbor, with a score of 76% on the f1-score test.