How does the world feel about cryptocurrency?
Announced Monday, media giant Thomson Reuters has launched a new bitcoin data feed designed to make it easier for investors know that very answer.
Included in the latest release of the company’s MarketPsych Indices product, a suite of investment tools analyzing everything from companies to sovereign bonds, the bitcoin sentiment data feed will use AI to analyze more than 400 sources of data, scouring news articles and social media posts in search of actionable insights.
Using metrics such as “greed” and “fear” both accredited an non-accredited investors can identify hedging opportunities or create buy-sell orders when a particular trait reaches a certain level or changes over a certain period of time.
And while Thomson Reuters has long tracked bitcoin prices, and last year made a push into applications of blockchain technology broadly, this week’s launch marks the firm’s formal entrance into products made specifically for cryptocurrency traders.
Still, Austin Burkett, the global head of Thomson Reuters quant and feeds division, acknowledged that the goal of the product is one all investors will find familiar.
He told CoinDesk:
“Our customers can use it to generate alpha. They can drive positive investment returns. They can use it to better balance the risk in their portfolios.”
AI for investing
Stepping back, the Thomson Reuters bitcoin data feed can also be seen as part of a larger trend among companies offering AI services, one that has found them seeking to leverage their existing tools for cryptocurrency investing.
Crowd sentiment firms Santiment and Token AI, for example, have emerged to target this use case, with a number of others entering as prices have risen in recent years.
At the core of all these investment insights products, however, are programs called Bayesian filters that apply lessons originally taught to computers by humans, according to Richard Peterson, author of “Trading on Sentiment: The Power of Minds over Markets.”
In interview, Peterson explained how sentences found in online source are first classified by human readers according