Hands-On Machine Learning for Algorithmic Trading

Key design choices

The design choices for neural networks are similar to other supervised learning models in some regards. For example, the neural network model output is usually dictated by the nature of the training labels and the type of ML problem they represent, such as regression, classification, or ranking. Based on the output, we need to select a cost function and an optimization algorithm to minimize this objective represented by the cost function.

Neural network-specific choices include the overall architecture, that is, the neural network's depth in terms of number of layers, and their relspective widths or number of nodes, as well as the design of connections between nodes of different layers. A key concern is how efficiently ...