Chris Porter Dow Jones Newswires - financial news for text mining
Jinfo Blog

Wednesday, 21st December 2016

By Chris Porter

Abstract

In addition to its Factiva service, Dow Jones provides real-time and archive feeds of its proprietary Dow Jones Newswires financial news offering, specifically for text mining purposes.

Item

Jinfo users who are investigating data mining projects may be well aware of the Factiva service from Dow Jones; but Dow Jones also has another set of services that may be of interest to them.


Machine-readable text feed 

Dow Jones Newswires is a global real-time news offering, serving financial professionals on third-party desktop terminals, on specialised financial websites and portals and in the company's own proprietary applications; and the Newswires content is also made available as a machine-readable text feed and archive, tailored to meet the needs of those undertaking analytics projects.

Options include ultra-low-latency delivery formats and feeds, for very high-speed algorithmic trading; and structured content archives, for historical analysis and back-testing.


Spotting unusual patterns 

Users of these Newswires feeds and archives include buy-side firms such as mutual funds, pension funds and insurance companies, who may be using them to help with asset allocation and investment strategies; and sell-side firms, who use them to help with market surveillance, looking for predictable or unusual spikes and patterns.

Financial regulators are also clients, and will typically be looking at how a financial news story relates to price movements - or alternatively they may be probing into situations where there is no relevant published news story, but still a big price movement to account for.


30 year archive 

The Dow Jones Newswires archive goes back over 30 years (though not all customers want the full archive); Dow Jones tells Jinfo that its attractions to users include the very structured, consistent writing style; the taxonomy and other metadata; and the precise timestamps, allowing easy correlation with "tick" data of market price movements.

« Blog