Python for Algorithmic Trading

Chapter 5. Predicting Market Movements with Machine Learning

Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th.

The Terminator (Terminator 2)

Recent years have seen tremendous progress in the areas of machine learning, deep learning, and artificial intelligence. The financial industry in general and algorithmic traders around the globe in particular also try to benefit from these technological advances.

This chapter introduces techniques from statistics, like linear regression, and from machine learning, like logistic regression, to predict future price movements based on past returns. It also illustrates the use of neural networks to predict stock market movements. This chapter, of course, cannot replace a thorough introduction to machine learning, but it can show, from a practitioner’s point of view, how to concretely apply certain techniques to the price prediction problem. For more details, refer to Hilpisch (2020).¹

This chapter covers the following types of trading strategies:

Linear regression-based strategies

Such strategies use linear regression to extrapolate a trend or to derive a financial instrument’s direction of future price movement.

Machine learning-based strategies

In algorithmic trading it is generally enough to predict the direction of movement for a financial instrument as opposed to the absolute magnitude of that movement. With this reasoning, the prediction problem basically boils down to a classification ...