The Expanding Data Universe: An Overview of Alternative Data Sources

For decades, traders have relied on a relatively small set of traditional data sources to make their decisions. These sources include things like stock prices, trading volumes, and company financial statements. However, in recent years, there has been an explosion in the availability of alternative data sources. These are data sources that are not traditionally used in financial analysis, but that can provide valuable insights into the performance of a company or a market.

Some of the most common alternative data sources include:

• News sentiment: The sentiment of news articles can be a good indicator of the future direction of a stock.
• Satellite imagery: Satellite imagery can be used to track a variety of economic activity, such as the number of cars in a retailer's parking lot, or the number of ships in a port.
• Social media: Social media can be a valuable source of information about consumer trends and public opinion.
• Credit card data: Credit card data can be used to track consumer spending patterns.

Streaming Alternative Data: The Power of WebSockets

Many alternative data sources are available as real-time streams. This means that the data is constantly being updated, and that it can be used to make trading decisions in real time. WebSockets are the ideal protocol for streaming alternative data. They provide a low-latency, bidirectional communication channel that can be used to deliver the data to the trader's desktop.

A Canonical Model for Unstructured Data: From Chaos to Order

One of the biggest challenges in working with alternative data is that it is often unstructured. This means that it is not organized in a way that is easy to process. For example, a news article is a block of text, and a satellite image is a grid of pixels. Before this data can be used in a trading strategy, it must be transformed into a structured format.

This can be done by using a variety of techniques, such as:

• Natural language processing (NLP): NLP can be used to extract structured information from text, such as the sentiment of a news article, or the names of the companies that are mentioned.
• Computer vision: Computer vision can be used to extract structured information from images, such as the number of cars in a parking lot, or the type of crops that are being grown in a field.

Once the data has been transformed into a structured format, it can be represented using a canonical data model. This will make it easier to work with the data and to combine it with other data sources.

Correlating with Market Data: Finding the Signal in the Noise

Once the alternative data has been transformed into a structured format, the next step is to correlate it with traditional market data. This will help you to identify any relationships between the alternative data and the performance of the market. For example, you might find that there is a positive correlation between the sentiment of news articles and the price of a stock.

It is important to be careful when correlating alternative data with market data. There is a risk of finding spurious correlations, which are correlations that are not statistically significant. It is therefore important to use a variety of statistical techniques to validate any correlations that you find.

Case Study: Building a Simple Trading Signal Based on a Real-Time, Normalized News Sentiment Feed

To illustrate how alternative data can be used to generate a trading signal, let's consider a simple case study. In this case study, we will build a trading signal based on a real-time, normalized news sentiment feed.

The first step is to subscribe to a news sentiment feed. There are a number of different providers that offer this type of feed. Once you have subscribed to a feed, you will need to normalize the data. This will involve transforming the data into a structured format and representing it using a canonical data model.

Once the data has been normalized, you can then use it to generate a trading signal. For example, you could generate a buy signal when the sentiment of the news is positive, and a sell signal when the sentiment is negative.

This is just a simple example, but it illustrates how alternative data can be used to generate a trading signal. By combining alternative data with traditional market data, it is possible to gain a significant edge over the competition.