Decision forests and factor models to improve investment strategies

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 three must-read academic papers that mix cutting-edge ML/DL with Asset Pricing & Quant Finance:

  • Forest through the Trees: Building Cross-Sections of Stock Returns

  • Application of Deep Learning for Factor Timing in Asset Management

  • Data-Driven Robust Statistical Arbitrage Strategies with Deep Neural Networks

• 💊 AI Essentials: The section on top AI & Quant Finance learning resources: Today, I'm thrilled to share an outstanding video titled "Algorithmic Trading – Machine Learning & Quant Strategies Course with Python." In this comprehensive course, you'll explore three cutting-edge trading strategies to enhance your financial toolkit. Dive into the Unsupervised Learning Trading Strategy, leverage the power of social media with the Twitter Sentiment Investing Strategy, and explore the Intraday Strategy with the GARCH model to capture both daily and intraday signals

• 🥐 Asset Pricing Insights: In this edition, I recommend the paper "Factor Timing with Portfolio Characteristics," which introduces a novel approach for timing factor portfolio returns by leveraging portfolio characteristics, advancing beyond traditional methods in asset pricing. This paper offers a fresh perspective on factor trading by demonstrating how portfolio characteristics can significantly enhance the predictability and performance of factor portfolios.

Today’s Sponsor

Copper is up 70%

Discover the Zonia Advantage

World Copper’s (OTC: WCUFF | TSX.V: WCU) Zonia project is leading the industry in next-generation copper production, and is poised to meet surging global copper demand. By leveraging simple metallurgy and proven SX-EW processing in a prime location, Zonia stands to achieve cash flow up to 4x faster than typical projects.

World Copper is one of the very few projects that can start producing before the end of this decade, just in time to capture the upcoming copper price surge.

LEARN MORE

(OTC: WCUFF | TSX.V: WCU)

AI-Finance Insights

“Forest through the Trees: Building Cross-Sections of Stock Returns”

🔊 How to mix decision forests and factor models to improve investment strategies? 👇

This paper introduces the Adaptive Pruning Trees (AP-Trees) methodology, a novel machine learning approach for asset pricing that enhances the selection and management of portfolios.

👉 AP-Trees leverage decision trees to create diversified portfolios by considering multiple firm-specific characteristics. This method ensures that each portfolio node traces back to economic fundamentals, offering a blend of flexibility and interpretability.

👉 Empirical analysis shows that portfolios constructed with AP-Trees outperform traditional triple-sorted portfolios. AP-Trees achieve significantly higher Sharpe ratios and alphas out-of-sample, indicating better risk-adjusted returns and more robust asset pricing performance.

👉 The study confirms that AP-Trees can efficiently capture interactions among multiple characteristics, which standard methods often overlook. This capability leads to more accurate and economically significant asset pricing models, highlighting the method's superiority in reflecting the true underlying economic risks.

“Application of Deep Learning for Factor Timing in Asset Management”

The paper explores the application of DL models to optimize factor timing in asset management, specifically focusing on the Conservative Minus Aggressive (CMA) factor premium.

👉 The study evaluates various models, including OLS linear regression, Ridge regression, Random Forest, and Fully-connected Neural Networks. The objective is to predict the CMA factor premium and use these predictions for factor timing. Flexible models like Random Forest and Neural Networks show superior performance in explaining the variance of the factor premium during the out-of-sample testing period, compared to linear models.

👉 The empirical analysis demonstrates that strategies based on Random Forest and Neural Network models achieve higher Sharpe ratios and better performance compared to traditional linear models. However, these advanced models also lead to more unstable optimal weights, resulting in higher transaction costs due to frequent rebalancing.

👉 The study addresses the impact of transaction costs, finding that higher fluctuation in weights, particularly in Random Forest and Neural Networks, can significantly erode returns. When proportional transaction costs are considered, even as low as 20 basis points, all models underperform the constant weighting scheme. To mitigate this, the paper suggests optimizing rebalancing frequency based on historical data, which can reduce transaction costs and improve overall returns.

👉 The research concludes that while deep learning models like Random Forest and Neural Networks provide better predictive power and factor timing performance, they are prone to higher transaction costs. Linear models, though less accurate in predictions, offer more stable and cost-effective solutions. This highlights the need to balance predictive accuracy with practical considerations like transaction costs in developing effective factor timing strategies.

“Data-Driven Robust Statistical Arbitrage Strategies with Deep Neural Networks”

The paper introduces a novel deep learning approach to identify robust statistical arbitrage strategies in financial markets. Unlike traditional methods that rely on cointegrated pairs of assets, this approach is model-free and data-driven, making it effective even in high-dimensional financial markets where classical pairs trading fails.

The methodology leverages deep neural networks to construct an ambiguity set of admissible probability measures derived from observed market data. This enables the development of trading strategies that are profitable under model uncertainty and various market conditions, including financial crises.

Empirical investigations demonstrate the robustness and profitability of these neural network-based strategies across different scenarios, highlighting their adaptability and superior performance compared to traditional statistical arbitrage methods.

Sponsor 👇

Learn AI in 5 Minutes a Day

AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.

Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.

Sign up with 1-Click

AI-Essentials

📹 Check out this engaging course on "Algorithmic Trading – Machine Learning & Quant Strategies with Python." In this comprehensive course, you'll explore three cutting-edge trading strategies to enhance your financial toolkit. 👇

Asset Pricing Insights

“Factor Timing with Portfolio Characteristics”

The paper introduces a novel approach for timing factor portfolio returns by leveraging portfolio characteristics, advancing beyond traditional methods in asset pricing.

👉 The proposed methodology utilizes dimension reduction techniques such as Principal Component Analysis (PCA) and Risk Premium PCA (RPPCA) to condense the predictor set and portfolio dimensions. Partial Least Squares (PLS) is employed to account for the covariance structure between predictors and forecasting targets.

👉 The study finds that characteristic-based models significantly outperform existing methods. These models achieve higher Sharpe ratios, with average monthly returns of up to 1.47% and annualized Sharpe ratios reaching 0.73. In comparison, the best benchmark model delivers 1.06% average monthly returns and a Sharpe ratio of 0.55. Notably, these strategies show no performance decay over time, a common issue with many individual anomalies.

👉 The combination of PCA, RPPCA, and PLS allows the models to efficiently capture interactions among multiple characteristics. This results in more accurate predictions and robust asset pricing performance. The study highlights the flexibility of these models to adapt to varying information signals over time, thus providing superior investment strategies.

Sponsor 👇

Get value stock insights free.

PayPal, Disney, and Nike recently dropped 50-80%.

• Are they undervalued?

• Can they recover?

• Read Value Investor Daily to find out.

We read hundreds of value stock ideas daily and send you the best.

Subscribe with one click.

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! 🌟🤖💼