Weekly Recap
Latest research on investing and trading
Welcome to this week's roundup of the latest investing research. Below is a carefully curated selection of highlights from the past week, with each title linking directly to its source for further reading.
Thank you for reading and don’t forget to hit the like button.
Commodities
Macroeconomic Conditions, Speculation, and Commodity Futures Returns (Adhikari and Putnam)
This paper tests the predictability of weekly commodity returns using a range of macroeconomic variables and measures of speculation derived from the Commitment of Traders report. The predictive power of speculation varies across time and commodity sectors, while the St. Louis Fed Financial Stress Index consistently emerges as a significant predictor across most commodities and periods. Higher financial stress is associated with lower commodity returns, with the strongest negative effects observed in the energy and precious metals sectors.
Crypto
Trump's Crypto Reserve Expansion: A Shift in U.S. Digital Asset Strategy (Krause)
Governments traditionally store reserves in gold, fiat currencies, and government bonds, but President Trump’s recent policy shift includes a broader mix of cryptocurrencies. The U.S. plans to diversify its crypto holdings beyond Bitcoin, adding assets like XRP, Cardano, and Solana. The paper explores the potential benefits of this strategy while also raising concerns such as possible market manipulation.
Equities
Decomposing Cross-Firm Return Predictability Using News-Wire Data (Zvonka)
Many studies explore return predictability among economically linked firms. This paper measures customer-supplier links using FactSet and finds that past customer returns on news days are better predictors of supplier returns than returns from non-news days. Investors can improve their strategies by focusing on news-driven returns, especially those related to earnings and dividends, to enhance performance.
Volatility of Price-Earnings Ratio and Return Predictability (Jiang and Li)
The predictive power of price-earnings (P/E) ratios has been studied extensively. However, this paper finds that the volatility of P/E ratios is a stronger predictor of future stock returns than both the level of P/E ratios and overall market volatility. It also predicts future market volatility and macroeconomic growth.
Decomposing Equity Returns: Earnings Growth vs. Multiple Expansion (Blitz)
Over the past decade, U.S. stocks have surged due to both strong earnings growth and rising valuations, leaving other markets and styles looking cheap in comparison. Breaking stock returns into earnings growth and multiple expansion reveals that small-cap and low-volatility stocks have solid profits but remain overlooked. At the same time, Emerging Markets struggle with weak earnings despite higher valuations. The paper highlights how market leadership shifts over time and suggests that well-diversified portfolios may benefit when undervalued segments regain favor.
Can we tame the factor zoo? The roles of alternative databases and extraction methods (Bessembinder, Burt, and Hrdlicka)
Many investors rely on factor models to explain stock returns, but the sheer number of factors, often called the "factor zoo", creates confusion. This paper shows that different methods for reducing the number of factors, as well as different datasets used, lead to widely varying results. No single approach consistently identifies the "right" factors. The findings suggest that conclusions from factor-based strategies can be highly sensitive to the choice of dataset and method, warranting caution in their application.
Asset Return Predictability: A Robust Assessment (Hau-Rueß, Ibragimov, and Min)
The seminal paper by Goyal and Welch (2008) found that common return predictors, such as dividend yields and interest rates, fail out-of-sample. However, this paper finds significant predictability by applying robust inference methods and theoretically motivated sign restrictions. For example, earnings-to-price and growth-adjusted valuation measures exhibit predictive power.
Factor Investing
A Conversation with Don Marcos (aka Marcos López de Prado) (Fabozzi)
Plenty of evidence suggests that factor investing has delivered unsatisfactory performance over the last decade, significantly underperforming in-sample results. In this recent interview, Dr. Marcos López de Prado discusses this issue. He stresses the importance of building factor models based on causality rather than relying on mere "associational" patterns in data.
Machine Learning and Large Language Models
The Role of Deep Learning in Financial Asset Management: A Systematic Review (Reis, Serra, and Gama)
This comprehensive review paper explores the literature on deep learning applications in asset management, highlighting emerging trends and the increasing use of alternative data, including sentiment analysis and ESG indicators.
A First Look at Financial Data Analysis Using ChatGPT-4o (Chou, Feng, Li, and Liu)
The authors examine how ChatGPT-4o performs in financial data analysis. While it can handle tasks like summarizing stock returns, estimating risk metrics, and conducting time-series analysis, it occasionally produces errors and lacks the judgment required for real-time investment decisions. The findings suggest that investors can use ChatGPT-4o as a complementary tool for data processing, but human oversight remains essential to ensure accuracy and interpreting market signals.
Can deep reinforcement learning beat 1/N (Kruthof and Muller)
Can deep reinforcement learning (DRL) outperform a simple equal-weighted strategy? This paper tests a DRL method across several datasets and finds that while it shows some market timing ability, it struggles to consistently beat an equal-weighted portfolio after accounting for trading costs due to high portfolio turnover.
Risk Management
Correlation: the most common mistakes made in Risk Management practice (Puccetti and Cagliani)
This paper highlights the shortcomings of the widely used Pearson correlation coefficient, such as its tendency to produce misleading results in non-linear relationships. Instead, it advocates for using rank correlation, which mitigates some of these issues.
Volatility
Forecasting realized volatility in the stock market: a path-dependent perspective (Liu, Fu, and Hong)
The heterogeneous autoregressive (HAR) model is a strong benchmark for volatility forecasting. This paper enhances the HAR model by incorporating path dependency, capturing past price and volatility trends. The results show that this path-dependent HAR model consistently outperforms the standard HAR model, demonstrating that accounting for historical price trajectories improves volatility prediction.
Blogs
Global FX management with systematic macro scores (Macrosynergy)
Learning to Rank (Quantitativo)
Batch Linear Regression via Bayesian Estimation (QuantStart)
Very.... slow... mean reversion .... and some thoughts on trading at different speeds (Rob Carver)
GitHub
skfolio - Python library for portfolio optimization built on top of scikit-learn
Medium
From Jupyter to Production: Deploying Machine Learning Models Like a Pro (Gupta)
Forecasting Volatility with the GARCH-VAR Model (Velasquez)
Podcasts
Scott Phillips - Finding Ugly Edges in Crypto Markets (Flirting with Models)
How Best to "Replicate" Managed Futures Returns ft. Katy Kaminski (Top Traders Unplugged)
Hendrik Bessembinder: Why It's So Hard to Beat the Market (Rational Reminder)
Last Week’s Most Popular Links
Refining ETF Asset Momentum Strategy (Pauchlyova and Vojtko)
The 10-year Expected Return of the S&P 500 in early 2025 (Philips and Kobor)
Spot-Futures Manipulations in Cryptocurrency Markets (Wang and Zhang)
Disclaimer: This newsletter is for informational and educational purposes only and should not be construed as investment advice. The author does not endorse any specific securities or investments mentioned. While information is gathered from sources believed to be reliable, there is no guarantee of its accuracy, completeness, or correctness.
This content does not offer personalized financial, legal, or investment advice and may not be suitable for your individual circumstances. Investing carries risks, and past performance does not guarantee future results. The author and affiliates may hold positions in securities discussed without prior notification.
The author is not affiliated with, sponsored by, or endorsed by any of the companies, organizations, or entities mentioned in this newsletter. Any references to specific companies or entities are for informational purposes only.
The brief summaries and descriptions of research papers and articles provided in this newsletter are the author's own interpretations of the findings and content. These summaries should not be considered as definitive or comprehensive representations of the original works. Readers are encouraged to refer to the original sources for complete and authoritative information.
This newsletter contains links to external websites and resources. The inclusion of these links does not imply endorsement of the content, products, services, or views expressed on these third-party sites. The author is not responsible for the accuracy, legality, or content of these external sites or for that of subsequent links. Users click on these links at their own risk.