2007 Sep 06 - Thu
News Analysis
I subscribe to DTN's IQFeed data
streams. (If you'd like to sign up, let me know I'll do a referral for you.) Anyway, in
addition to the usual equity, futures, and options feeds, they have a news feed. Each feed
entry has a media source indicator, a headline, a list of associated symbols, and a index
number for obtaining the story content.
I thought it might be an interesting project to process each incoming message for its
symbol list and do some sort of key word analysis to see if one can get a 'mood' of the
article. This might provide some interesting trading ideas for the day.
I don't have the
time to do it right now, but am recording my thoughts so I can
come back to it a little later.
Two recent articles by Paul C. Tetlock in the The Journal of Finance, one in the June 2007 issue titled "Giving Content
to Investor Sentiment: The Role of Media in the Stock Market", and one in an upcoming issue
called "More Than Words: Quantifying Language to Measure Firms. Fundamentals", got me
thinking about this again.
One of the articles pointed to the General Inquirer, no, not a racy tabloid but a "a computer-assisted approach for content analyses
of textual data". Although GI references an application useful for researches, I think the
interesting content resides with the spreadsheet of categorized words they have. These
words can be used to classify the 'mood' of processed text.
The site also points to a book called "The Content Analysis Guidebook" by Kimberly A.
Neuendorf as one that might shed further background on the concept. A while ago, I was
taking a look at content anlysis from a different perspective, something akin to classifying
market analysis and trading blogs. Some additional book references are linked below.
An application called Yoshikoder is an
already built application that can take the GI word lists and process portions of text and
produce analysis summaries.
A brief web search brought up a couple of blogs that show some perspective on how to put
analysis into perspective:
Some 'possibly' related books:
|
