The notion that text-based data is useful for trading financial markets is not an unusual concept. After all, news has been a major driver of trader behavior and prices for centuries. What has changed in recent years is the sheer quantity of text-based data that a trader might need to look at, in particular driven by the advent of the web. There is simply too much text for any human to read and interpret. We need to turn to machines to help us extract value from this huge quantity of text for us to use in the investment process.
In this chapter, we begin by exploring how to read web data. We then give many use cases for text from an investor viewpoint. We look at social media and show how it can be used to understand ideas such as market sentiment and to help forecast US change in nonfarm payrolls. Later, we will focus on newswire data and develop systematic trading rules by using it for FX markets. We will also discuss how to aggregate Fed communications and apply NLP to it to understand the movement in US Treasury yields. Lastly, we will talk about making estimates for CPI using web-sourced data from online retailers.
15.2. COLLECTING WEB DATA
The web was invented in 1989 by Tim Berners-Lee while he was working at CERN. Obviously today, over 30 years later, the amount of content available on the web has mushroomed. The web can encompass content such as news, social media, blogs, corporate data, and so on, but ...