Machine Learning for the Markets

This two-day course is a unique opportunity for delegates to learn the main machine learning algorithms as well as how they can be used to tackle problems in the financial markets. The course covers a wide range of techniques, from classification, clustering, dimensionality reduction to regime switching models and structural analysis of time series. Each delegate will be equipped with a PC to perform the R exercises on each topic using financial data.

Introduction to R and Machine Learning

  • Overview of machine learning and associated fields
  • Brief description of main R commands
  • Overview of main R packages for Machine Learning Analysis

Clustering / Unsupervised Learning

  • Overview of main clustering algorithms (linear and non-linear K-means clustering)
  • R Exercise: application of K-Means clustering to volatility

Classification / Supervised Learning

  • Main classification algorithms: SVMs, decision trees, neural networks, random forests
  • R Exercise: application of classification algorithms to directional prediction

Function Approximation / Supervised Learning

  • Overview of algorithms (support vector regression, linear regression, neural networks)
  • R exercise: application of above algorithms to return forecasting

Genetic Algorithms

  • Overview of genetic algorithms
  • R exercise: application of genetic algorithms to optmisation of technical strategies

Dimensionality Reduction / Variable Selection

  • Overview of main dimensionality reduction algorithms (e.g. PCA, Sparse PCA)
  • Methods to select relevant variables for modelling
  • R exercise: application of PCA to systemic risk

Regime Switching Models

  • Overview of main regime switching models (e.g.: Markov Switching Models)
  • R Exercise: application of markov models to volatility regime switching