There are several helpful uses of unsupervised learning that can be applied to algorithmic trading, including the following:
- Grouping together securities with similar risk and return characteristics (see hierarchical risk parity in this chapter (which looks at portfolio optimization))
- Finding a small number of risk factors driving the performance of a much larger number of securities
- Identifying trading and price patterns that differ systematically and may pose higher risks
- Identifying latent topics in a body of documents (for example, earnings call transcripts) that comprise the most important aspects of those documents
At a high level, these applications rely on methods to identify clusters and methods to reduce the dimensionality of the data.