Quant Hedge Fund Winton to Offer Big Data Services to OutsidersBy
Hivemind collects, cleans and enhances messy data sets
Unit adds new revenue source for the $30 billion hedge fund
Winton Group, the $30 billion quantitative hedge-fund firm run by billionaire David Harding, plans to offer the services of a data unit that helps power its computer-driven algorithms to outsiders.
Hivemind, which was part of Winton’s research division before being incorporated as a separate company in July, is preparing to license its technology and provide consultancy to other firms to help solve their data problems. The unit collects, cleans and enriches complex datasets, and has been offering the services to artificial intelligence firms in which Winton is invested.
“We are already in talks with top-tier investment banks on a broad set of uses for Hivemind, ranging from modernizing their sell-side research to improving customer data and managing over-the-counter trade documentation," Henrik Grunditz, head of business development at the unit, said in an interview.
Hedge funds such as Winton are at the forefront of extracting meaningful information from ever-increasing data produced worldwide -- from cellphones spitting out geographical locations and web searches to modern satellites taking high resolution images of parking lots or oil rigs -- as they hunt for fresh trading ideas. Sharing that expertise for a fee adds another revenue source as firms fight sluggish returns and dwindling fees.
An estimated 90 percent of the data that exists today has been created in the last two years alone, JPMorgan Chase & Co. wrote in a report in May. But the data deluge is hardly useful for machines in its current form as most of it is unstructured and designed for human consumption in the form of text documents, audio and images.
“It’s not nice, neat tables that you can run correlations on," said Daniel Mitchell, who runs a team of data scientists for Winton. “Hivemind is a technology that can build useful and accurate datasets out of those unstructured data sources and do it quickly."
About three quarters of Winton’s current research projects have used Hivemind, cutting the firm’s reliance on external data providers. The unit created 20 unique data sets in the last year alone.
Data sets include M&A transactions by S&P 500 companies going back to the 1960s and monthly data on movement of commodities and other goods through the Suez Canal from 1870 to 1956.
Winton doesn’t plan to sell its proprietary data sets, though it could make some of them available to non-competing firms at a later stage, the company said.
Bloomberg LP, the parent of Bloomberg News, also provides data and analytics to financial services companies.