Machine Learning for Smart Beta

Overview

In this 4 week course, you will learn about Smart Beta products. Smart betas products have the characteristics of both passive investment(having predetermined rules) and active investments(allows for factor investment). We will walk through the creation mechanisms behind different smart beta products and recreate some of them using R programming. Then we will apply machine learning methods. Data processing, overfitting prevention techniques will be covered. Finally we will try to create an improved multi-factor model using CART, bagging, boosting and ensemble methods. Students are expected to have listened to my first and second course 'The Fundamental of Data-Driven Investment' and 'Using R for Regression and Machine Learning in Investment', or having equivalent knowledge in investment concepts and a firm grasp on R programming.

Syllabus

Week 1

Building on the concepts learned in previous courses 'The Fundamental of Data-Driven Investment' and 'Using R for Regression and Machine Learning in Investment', this course will cover 'Smart beta'. Smart betas products have the characteristics of both passive investment(having predetermined rules) and active investments(allows for factor investment). Smart beta products' investment mechanisms are open to the public, so we will recreate a MSCI smart beta product in R. Follow along the step-by-step reconstruction of the MSCI Enhanced Value Index and create your own smart beta portfolio.

Week 2

In order to effectively utilize machine learning in investment, it is important to understand the various characteristics of data. This module covers how to check the prediction accuracy of a machine learning model and prevent overfitting. Get hands on experience in R to manipulate data into a form suitable for machine learning models from regression models to classification trees.

Week 3

The asset selection method based on a score derived from a benchmark index has the problem that the selected assets do not reliably capture underlying information. To solve this problem, a non-traditional method, namely machine learning is used to create an improved multi-factor approach. Familiarize yourself with CART(Classification and Regression Tree), bagging, boosting and ensemble methods to enhance your smart beta portfolio in R.

Week 4

In this final module, we wrap up the discussion by creating a multifactor model applying all the knowledge we have learned so far. Investors have taken a steady interest in multifactor models that take into account the cyclicality of factors. Further, we expand the discussion into the use of factors in bond investment and a new method of active factor allocation.