Forecasting Option Returns with News and Machine Learning

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:

  • Forecasting Option Returns with News.

  • Visualizing Earnings to Predict Post-Earnings Announcement Drift: A Deep Learning Approach.

• 💊 AI Essentials: The section on top AI & Quant Finance learning resources: Today, I'm sharing a video perfect for total beginners—a comprehensive introduction to Reinforcement Learning. This video walks you through the basics, providing an accessible foundation for understanding key RL concepts. If you're new to AI and want to explore Reinforcement Learning from the ground up, this is an excellent starting point to unlock its potential.

• 🥐 Asset Pricing Insights: In this edition, I introduce "New Investment Ideas! Optimizing Sharpe Ratio with Multi-Armed Bandits"—a paper that explores how optimizing risk-adjusted returns can lead to better decision-making in portfolio management. The authors present a novel approach to optimizing the Sharpe Ratio (SR) using multi-armed bandit (MAB) algorithms, offering fresh insights into risk-adjusted decision-making.

Today’s Sponsor

Innovative startup makes window automation accessible for all

RYSE SmartShades let you retrofit existing blinds in less than five minutes.

You can set schedules to open at sunrise and close at sunset.

All this comes with seamless Google Home or Amazon Alexa integrations

Take a look at their products yourself.

“Forecasting Option Returns with News”

💊 This paper explores the predictive power of news media for equity option returns using machine learning. By leveraging text-based signals from news articles, the authors provide a novel method to forecast delta-hedged option returns.

Key contributions and findings include:

👉 Text-based signals derived from news articles are shown to significantly predict delta-hedged equity option returns, even after controlling for existing option return predictors.

👉 The authors use support vector regression (SVR) to extract textual signals, demonstrating robust predictive power across different text representations and machine learning methods, including elastic net, random forest, and neural networks.

👉 News coverage is found to contain valuable information about future changes in stock return volatilities, which serves as a key driver of option return predictability.

👉 The study highlights the usefulness of news data in providing unique insights into the option market, outperforming traditional dictionary-based sentiment measures in predicting option returns.

“Visualizing Earnings to Predict Post-Earnings Announcement Drift: A Deep Learning Approach”

👉 This paper investigates the predictive power of visualized earnings on post-earnings announcement drift, employing deep learning to extract crucial features from earnings data. Specifically, the authors transform firms' quarterly earnings into bar charts and use a Convolutional Neural Network (CNN) to identify features most indicative of post-earnings announcement performance.

👉 The study finds that CNN-extracted features significantly predict post-earnings announcement returns, outperforming traditional predictors and maintaining stability over time. The features capture insights that typical risk controls and return anomalies do not fully explain.

👉 The CNN model assigns probabilities to different investment actions ("buy", "hold", "sell") based on earnings data visualization, demonstrating that firms with higher "buy" probabilities consistently outperform others, with returns exceeding those of existing drift predictors.

👉 The model helps uncover overlooked factors by investors, showing that visualized earnings data offers valuable predictive information for future earnings growth that the market fails to incorporate fully, resulting in exploitable opportunities.

👉The results are robust and comparable to machine learning techniques, highlighting the importance of visual representation in forecasting.

Sponsor 👇

Better PR with minimal effort: Let AI write your articles

• Generate high-quality articles in seconds - SEO-optimized, plagiarism & fact-checked

• Be featured for free by journalists looking for credible sources and build authority

• Get your articles indexed and ranked directly in Google News

• Distribute your content to top magazines with a single click

• Forget about ChatGPT: Manage, publish and track your PR efforts in one place

Get great PR fast:

Try Pressmaster.ai now for free

AI-Essentials

In this video, we're diving into the world of Reinforcement Learning! We'll start with the basics and gradually explore more advanced topics like Deep Reinforcement Learning and how to implement these concepts using PyTorch.

Asset Pricing Insights

“Optimizing Sharpe Ratio with Multi-Armed Bandits”

💊 New Investment Ideas! "Optimizing Sharpe Ratio with Multi-Armed Bandits" How optimizing risk-adjusted returns leads to better decision-making in portfolio management. Keep reading!🔻

👉 This paper introduces a novel approach to optimizing the Sharpe Ratio (SR) using multi-armed bandit (MAB) algorithms for risk-adjusted decision-making. The authors propose maximizing a regularized variant of the SR—Regularized Square Sharpe Ratio (RSSR)—to overcome challenges in deriving online algorithms for SR optimization.

👉 The UCB-RSSR algorithm is introduced for regret minimization (RM) in MAB settings, showing improved performance in maximizing RSSR with a regret bound that scales as O(log n) for the two-armed bandit case.

👉 The authors also propose SHVV, SHSR, and SuRSR algorithms for best arm identification (BAI) in a fixed budget setting, aimed at selecting the arm with the highest variance or SR. These methods provide upper bounds on the probability of error, demonstrating superior efficiency in identifying the best arm.

👉 Empirical results indicate that UCB-RSSR outperforms existing SR optimization algorithms, including U-UCB and benchmarks like GRA-UCB and MVTS, across various distributions such as uniform, truncated Gaussian, and gamma.

👉 The proposed framework highlights practical applications in risk-aware portfolio management, providing a powerful tool for maximizing returns while managing volatility.

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! 🌟🤖💼