Dynamic Asset Allocation with Asset-Specific Regime Forecasts

Hi! Here's Iván with this week's exciting newsletter, brimming with insights and discoveries on building robust investment strategies and risk models using Machine Learning.

In this edition, I am presenting the following sections:

• 🕹️ AI-Finance Insights: I summarize two must-read academic papers that mix cutting-edge ML/DL with Asset Pricing & Quant Finance:

  • Dynamic Asset Allocation with Asset-Specific Regime Forecasts

  • Can Machine Learning Methods Predict Beta?

• 💊 AI Essentials: The section on top AI & Quant Finance learning resources: Today, I’m introducing a practical tutorial on building a machine learning model to predict stock market prices, perfect for mastering key concepts in AI, including neural networks and deploying real-time applications using Python.

• 🥐 Asset Pricing Insights: In this edition, I introduce an great paper that explores a deep learning strategy leveraging limit order book data to forecast multi-horizon stock returns, providing a more effective method for optimizing high-frequency trading, especially when conventional techniques underperform.

Today’s Sponsor

Surmount democratizes investing, making advanced strategies accessible to everyone. Craft, analyze, execute strategies effortlessly. Total control, precision, and confidence await.

Surmount provides the tools and insights you need to succeed.

Start investing through data!

“Dynamic Asset Allocation with Asset-Specific Regime Forecasts”

Discover how ML can improve short-term return predictions with a focus on stability. Keep reading! 🔻

This paper presents a cutting-edge hybrid strategy for asset allocation that uses machine learning to enhance predictions and optimize portfolios.

The main contributions and findings are as follows:

👉 Hybrid Regime Forecasting: This new framework applies both unsupervised and supervised learning techniques to identify and predict bullish or bearish market regimes for individual assets, improving accuracy and asset-specific insights.

👉 Customized Forecasting: By moving beyond broad economic regimes and focusing on specific asset dynamics, the method enhances the signal-to-noise ratio and allows for more precise portfolio adjustments in response to market shifts.

👉 Dynamic Asset Allocation: Integrating these asset-specific regime forecasts into the Markowitz mean-variance optimization model, this approach helps create optimal allocation weights, balancing return and risk more effectively.

👉 Proven Performance: Empirical studies show that this model outperforms traditional strategies like minimum-variance and naive-diversified portfolios, demonstrating stronger returns and lower risk across various asset classes.

👉 Adaptable to Market Volatility: This framework offers investors a dynamic tool for managing portfolios in fast-changing markets, making informed decisions to capture opportunities while mitigating risk.

“Can Machine Learning Methods Predict Beta?”

How ML can help in beta estimation for non-traded assets. Keep reading! 🔻

This paper presents a new approach to beta estimation, using ML to vastly improve accuracy for private and non-traded firms.

The main contributions and findings are as follows:

👉 Machine Learning Beta Estimation: The paper compares traditional Comparable Company Analysis (CCA) with five machine learning (ML) algorithms for estimating beta, including decision tree ensembles and neural networks.

👉 Dramatic Error Reduction: The ML models significantly reduce mean absolute error (MAE) by up to 42% compared to CCA, with particularly strong performance for smaller, younger firms—those that typically differ most from their peer groups.

👉 Outperformance in Non-Traded Asset Valuation: ML methods outperform traditional accounting betas and provide better estimates for firms with unconventional capital structures, making them ideal for private firm valuations.

👉 Robustness Across Models: The study proves the robustness of ML results even with fewer predictors and demonstrates that the Light GBM method performs best, consistently improving forecast accuracy across different firm sizes and ages.

Sponsor 👇

Transform the way you run your business using AI (Extended Labour day Sale)💰

Imagine a future where your business runs like a well-oiled machine, effortlessly growing and thriving while you focus on what truly matters.
This isn't a dream—it's the power of AI, and it's within your reach.

Join this AI Business Growth & Strategy Masterclass and discover how to revolutionize your approach to business.
In just 4 hours, you’ll gain the tools, insights, and strategies to not just survive, but dominate your market.

What You’ll Experience: 
🌟 Discover AI techniques that give you a competitive edge
💡 Learn how to pivot your business model for unstoppable growth
💼 Develop AI-driven strategies that turn challenges into opportunities
⏰ Free up your time and energy by automating the mundane, focusing on what you love

🗓️ Tomorrow | ⏱️ 10 AM EST

This is more than just a workshop—it's a turning point.
The first 100 to register get in for FREE. Don’t miss the chance to change your business trajectory forever.

Sign up here to save your seat! 👈

AI-Essentials

In this video, you'll learn how to build a complete machine learning project in Python to predict stock market prices, from data preprocessing to model deployment as a real-time web application. The tutorial covers creating a neural network using TensorFlow Keras with Dense, Dropout, and LSTM layers, and is explained in a beginner-friendly way.

Asset Pricing Insights

“Improving Alpha Term Structures from Order Book Data”

The paper presents a nice deep learning strategy that leverages limit order book data to forecast multi-horizon stock returns and optimize high-frequency trading.

The main contributions and findings are as follows:

👉 The framework explores deep learning models like LSTM and Seq2Seq to forecast alpha term structures—future stock returns across multiple time horizons—using raw order book data.

👉 The study evaluates four DL models: a simple LSTM, multi-head LSTM, LSTM Seq2Seq, and LSTM Seq2Seq with attention. Surprisingly, the simple LSTM consistently outperforms more complex models, including those with attention mechanisms.

👉 Smaller batch sizes (64-256) significantly boost forecasting performance, while increasing depth slightly improves accuracy. However, increasing model width degrades results.

👉 Incorporating time features such as time of day and the interval between order book updates significantly enhances predictive power, especially towards the end of the trading day.

👉 The study confirms that more stable order books lead to stronger forecasting performance, offering valuable insights into how market conditions affect high-frequency return predictions.

If you're enjoying our newsletter and want to support us, please recommend it to anyone you know who's interested in AI and Finance. Your referrals are the biggest compliment and help us grow! 🌟🤖💼