Background

Articles from reputable sources about companies often have immediate impacts on that company’s stock. Investors will act based on information that they have gathered through their environment, either through discussions with co-workers or friends, reading of media articles, or a mathematical algorithm. The efficient-market hypothesis states that the asset price of a stock fully reflects all available information. Thus it should be theoretically impossible to consistently beat the market on a risk-adjusted basis because everyone has access to the same information and will make informed decisions based on the same information. However, with the dawn of new technology and our ability to use that technology to investigate general stock trends, savvy investors will use the most advanced techniques to attempt to “beat the market”. One demonstration that is not generally available to the public is the use of machine learning algorithms to determine the sentiment of an article that is released.

The stock price of a company is a reflection of how much investors think the company is and will be worth. A primary agent through which the public values a company is the news organizations that divulge information about the company's ROI, potential company direction, CEO statements regarding the future direction of the company, purchases, employee status, etc… When positive information surrounds a company, the value of that stock will increase, reflecting the investor’s belief that the company is worth more. However, if negative information surrounds a company, the value of that stock will diminish. Being able to accurately and quickly determine the direction of public sentiment surrounding about a company has very valuable implications in terms investment profit. Thus, with the predictive power of artificial intelligence, big data and computing, an investor with access to all of the media articles about a given company could, in theory, accurately predict the company’s stock performance. If the sentiment across all the articles is in general positive, then that investor could safely assume that the public will read the articles, perceive the company as doing well and trade accordingly, inflating the value of the stock and conversely for the opposite case.

How does one get access to all of this information? Many news organizations now release RSS feeds to their websites. An RSS (rich-site- summary) is an XML formatted document containing all the articles for a given news organization. The information is organized based on an article, and contains metadata on the title, date, url, etc. Subscribing to these RSS feeds allows a user to have instant access to the articles of a given news organization. An efficient program can then constantly run and check for any changes to these feeds. Being able to react instantaneously upon the release of this information then creates a potential advantage in the market.