Machine Learning and Stochastic Control for Algorithmic Trading
Algorithmic trading refers to the automatic trading of assets using a predefined set of rules. These rules can be motivated by financial insights, and/or mathematical and statistical analysis of assets. Price, order-flow, and posted liquidity are often factors in determining how to trade. This minicourse will look at algorithmic trading problems using methods of stochastic control and machine learning. We will first study how data behave, by examining machine learning methods to classify and model data, such as Bayes classifier, multi-logistic regression, support vector machines, and hidden Markov models. We will study how to pose stochastic control problems, and show how certain class of problems including pairs trading and statistical arbitrage can be solved mathematically. Finally, we will study reinforcement learning methods (which attempt to learn optimal behavior without putting model assumptions on the data), and combine them with partially observed stochastic control problems to include latent states.
– Limit Order Books
– Optimal Execution
– Hidden Markov Models
– Functional Principal Component Analysis
– Reinforcement Learning
This short course is partly based off of my book Algorithmic and High-Frequency Trading.
|Algorithmic and High Frequency Trading,
Cambridge University Press, now available!
Click here for the book website where you can find data, code and other materials related to the book.
Notes and Videos
|1||Introduction to Limit Order Books, and some empirical facts||RiO-Limit-Order-Book|
|2||Functional Data Analysis and Order-Flow||RiO-Order-Flow|
|3||Classification: Bayes classifier, simple Bayes, and Multi-Class Logistic||RiO-Classification|
|5||Hidden Markov Models||RiO-HMM|
|7||Trading with Latent Alpha||RiO-LatentAlpha|