Webinars This Blog

Portfolio Assets Allocation with ML and Optimization for Dividend Stocks | Algo Trading Projects

3 min read

Complete recording

About the Presentations

EPAT project presentations by two of our esteemed EPAT alumni - “Portfolio Assets Allocation: A practical and scalable framework for Machine Learning Development” by Raimondo Marino from Milan, Italy and “Portfolio Optimization for Dividend Stocks” by Kurt Selleslagh from Singapore.

Project 1: Portfolio Assets Allocation: A practical and scalable framework for Machine Learning Development

A framework for portfolio allocation for both traditional and Machine learning model based was developed to design and evaluate different portfolio techniques such as Equal Weighted Portfolio (EWP), Inverse Volatility Portfolio (IVP), Hierarchical Risk Parity (HRP) and Hierarchical Equal Risk Contribution (HERC).

As is often the case, ML techniques outperform traditional ones when allocating weights to different assets, therefore, providing superior performances and greater risk reduction through better diversification. The idea was to design a market neutral (long/short) portfolio of assets to be rebalanced periodically choosing different assets during every rebalance.

The criteria used for selecting assets during the rebalance for the long/short portfolio was the Cross-Sectional Momentum trading strategy. Then, using different portfolio techniques, an unlevered and normalized weight allocation was performed on the asset universe. This process was followed by a vectorized backtest whose objective was to assess the performances of the different techniques.

Lastly, a Montecarlo simulation was conducted on the best portfolio to properly size the amount of money to be invested during the periodic portfolio rebalancing. Specifically, two metrics were simulated: the Terminal Wealth and the Max DrawDown. Moreover, there was no parameters optimization since HRP and HERC are ML techniques in the realm of Unsupervised Learning field.

This way we prevented the risk of overfitting. Everything was accomplished through python classes that allow for a fast and smooth transition from Development to Production environments. Another technique used for improving the code efficiency was the use of the multiprocessing python module that allows the use of parallel computing (multiple CPUs) when performing backtesting on thousands of different parameter combinations.

Slide deck

Project 2: Portfolio Optimization for Dividend Stocks

This project extends the concept of portfolio optimization to dividend-paying stocks. However, we are not only going to optimize the net price return, but also - and more importantly - the dividend yield.

We seek to achieve an attractive regular dividend income stream (4% or higher) while at the same time obtaining a positive price return (no loss in invested capital).

Through optimal portfolio selection, we seek to outperform the benchmark index (or ETF) for dividend-paying stocks (and REITS), both in terms of dividend yield and price return.

About the Presenters

Kurt Selleslagh (Project Manager at Hong Kong Exchanges and Clearing Limited - HKEX)

Kurt Selleslagh pic

Kurt has more than 27 years of experience in the financial markets. He has held various positions like IT Project Manager, Business Project Manager, Quality Assurance test lead, Scrum Master, Business Analyst, and Consultant in his career with reputed exchanges and institutions like Hong Kong Exchanges and Clearing Limited, Standard Chartered Bank, NYSE Euronext, and The Capital Markets Company.

He has extensive knowledge of financial markets, products (especially derivatives), asset classes (equity, FX, commodities, interest rate), systems and regulatory environment. Kurt possesses deep domain knowledge of front office, electronic trading, post-trade processing (middle- and back-office), risk management, CCP clearing, OTC Clearing, quantitative methods for trading and risk management, and algorithmic trading.

Kurt holds an MSc in Financial Engineering degree from the Nanyang Technological University in Singapore and also holds numerous certifications in Fintech domains like Blockchain, Project Management, Cloud Infrastructure, Python Programming and Algorithmic Trading. He is passionate about driving and delivering new initiatives – infrastructure solutions, new products, platform implementation and process improvement - within the financial services industry.

Raimondo Marino (Artificial Intelligence & Big Data Engineer - Freelance)

Raimondo Marino pic

Raimondo Marino is a professional freelance working as an Artificial intelligence Engineer for Italian Small and Medium Companies. Through AI applications, he comes up with end to end solutions (from Development to Production using cloud services) for different corporate functions within a company: Marketing, HR, Sales, Production, etc.

He is also a passionate (literally “mad”) quant-trader who believes in the efficacy of combining Machine Learning techniques with Statistics and Probability to design superior trading systems. Previously He worked for over 20 years in the Telecom industry mainly in the Business Marketing offering and base Management.

Raimondo holds a Master’s degree in Electronic Engineering from University of Naples (Italy), a Master of Business Administration (MBA) from SDA Bocconi of Milan (Italy) and a Postgraduate in Machine Learning and Artificial Intelligence from Columbia Business Executive Education (USA). He is 53 years old married with a 7-year-old daughter.

This event was conducted on:
Tuesday, December 13, 2022
8:30 AM ET | 7:00 PM IST | 9:30 PM SGT

Algorithmic Trading Course for Free