Machine learning — the practice of teaching machines to make decisions using existing data — can do a lot of things. One academic researcher has devised a way to use it to predict the price of the Bitcoin virtual currency.
That might not sound like a big deal at first. But think about it for a moment. With the system, you could figure out when to sell the remaining Bitcoins you have in order to minimize your loss or maximize your gain. Or you could find out the right time to buy.
“Can we explain the price variation in terms of factors related to the human world? We have not spent a lot of time doing that,” Devavrat Shah, associate professor of electrical engineering and computing science at the Massachusetts Institute of Technology (MIT), told a writer with the university’s news office for an article today. “But I can show you it works. Give me your money and I’d be happy to invest it for you.”
That’s some gumption right there. And it shows that machine learning can go beyond detecting astronomical phenomena in images, recognizing products and people in text, and making movie recommendations. Machine learning could actually help brave Bitcoin investors make money.
“The strategy experimented in this paper holds [a] minimal position — at most 1 Bitcoin (+ or −). This leads to nearly doubling of investment in 50 days,” write Shah and his former colleague Kang Zhang in a recent paper.
Learn about their Bitcoin trading strategy, which entails making a trade based on Bitcoin’s current position, in the full academic paper (PDF).