Finding Alpha with Financial Machine Learning and Alternative Data
Alternative data can be defined as any non-obvious leading indicator of financial performance. Nowadays, people talk about satellite imagery and social media analysis when they talk about alternative data. Before the dawn of the internet, it existed in different forms.
For example, if you wanted to know how many customers a hardware store has in any given day, what would be stopping you from standing outside their front door and counting the number of customers that walk in? Or, following scuttlebutt method, why not ask the customers how they like the hardware store compared to others in the area?
All of these questions and modes of analysis produce data that may or may not be useful for predicting price movements of financial securities. Financial machine learning can convert these streams of raw data into signals that can be used to make predictions about price movements.
What is financial machine learning?
Financial machine learning is not plug-and-play. It should be considered an entirely different domain of thought from traditional machine learning or deep learning. Quite a lot of work needs to be done to transform streams of disparate data series into actionable investing signals.
Access to valuable alternative data does not guarantee good trading performance. For example, imagine you have access to insider information on a company. Specifically, you have access to the revenue figures for the company 15 days prior to the official public announcement, every quarter, with perfect accuracy. Access to this type of data still does not guarantee that you will make money trading that company's stock. You still have to answer the following questions.
- Is the revenue figure a positive or negative surprise?
- By how much and in which direction will the announcement of the revenue figure move the price of the stock?
- What other fundamental factors affect have an effect on the announcement of the revenue figure? For example, did the company drastically increase revenue because they recently completed an LBO of a competing company?
In other words, access to unique information does not confer an absolute advantage. It confers a potential advantage. Financial machine learning can allow us to extract the valuable signals from alternative data streams that can not otherwise be discovered by manual inspection.
We can help you development both alternative data and financial machine learning solutions to improve your investing approach.