An MIT professor and his student have written a paper describing the use of an algorithm to predict the fluctuating price of Bitcoin.
Bitcoin is a digital currency, which is collected in exchange for lending processing power or bought from providers.
The paper, by Devavrat Shah and Kang Zhang, explored whether using historical Bitcoin data from currency exchange company OKCoin could help predict the future price of the currency when the Bayesian regression algorithm is applied.
“Instead of making subjective assumptions about the shape of patterns, we simply take the historical data and plug it into our predictive model to see what emerges,” Shah told MIT News.
“We were also intrigued by the challenge of predicting a currency that has seen its prices fluctuate regularly in the past few years.”
More on Bitcoin
- Bitcoin exchanges maturing, but Bitcoin security still a concern
- Compliance for Bitcoin, virtual currencies, storage and backup
- Is Bitcoin safe? The truth about Bitcoin security and crypto-currency
- Reed.co.uk and Bitcoin – using agile to drive innovation
The study uses 200 million points of data from between February and July 2014 to map the future price of Bitcoins in 10 second intervals over the time periods of 30, 60 and 120 minutes. These predictive prices can then be used to determine whether to buy or sell Bitcoins, or do nothing.
This conversion rate can make it difficult to implement a system allowing Bitcoin as payment. Mark Ridley from Reed.co.uk explained how the recruitment company implemented Bitcoin into its payment system by converting the price of the currency to pounds at the point of transaction.
"It might not be that Bitcoin is the actual currency that wins out in the long run, but certainly as a proof of the theory behind that type of currency, it is incredibly powerful,” he said.
Shah’s MIT paper explains the Bayesian regression algorithm has been used in previous studies to predict trending topics on Twitter.