Deep Learning for Finance

Chapter 11. Advanced Techniques and Strategies

By now, you should have a solid understanding of deep learning algorithms and how to develop a model to predict time series data. Even though this is just a first step toward deploying a profitable algorithm, you should know that you have come a long way since the beginning of the book. This chapter is divided into independent sections that discuss interesting ways of applying a few advanced deep learning techniques and methods for time series prediction and to enhance the process.

Using COT Data to Predict Long-Term Trends

The Commitments of Traders (COT) report is a weekly publication released by the US Commodity Futures Trading Commission (CFTC). It provides information on the positions held by various market participants in futures markets. The report is based on data collected from futures exchanges, including the Chicago Mercantile Exchange (CME) and the Intercontinental Exchange (ICE). The COT report categorizes traders into three main groups:

Commercial traders (also referred to as dealers or hedgers)

These are typically companies that use the futures market to hedge their main business activities. For example, a grain producer may use futures contracts to protect against price fluctuations in the agricultural market. Their positions are generally negatively correlated to the underlying market.

Noncommercial traders (also referred to as funds or leveraged money)

This group consists of large speculators, such as hedge funds ...