Weekly Research Recap
Latest research on investing and trading
It's time once again to explore some of the most compelling investing research from the past week. Below, you'll find a hand-picked selection of recent papers, each linked directly to the original source for further reading.
Thanks for your ongoing support! If you enjoy this content, please consider hitting the like button and subscribing if you haven’t yet.
Behavioral Finance
The Behavioural Characteristics of Fund Managers (Harris and Mazibas)
Most assume professional fund managers are immune to behavioral biases, but this paper shows otherwise. By analyzing actual returns from nearly 200,000 funds, the authors find that managers consistently display psychological patterns like loss aversion and distorted probability weighting. Interestingly, they’re less loss-averse than retail investors but more prone to overweighing rare outcomes. Fund managers, it seems, are guided by human psychology just like retail investors, though in distinct ways.
Bonds
Exploring emerging markets debt: Bond voyage? (Giesta and Swinkels)
Emerging market debt has grown rapidly, but its value to investors has shifted over time. In the 2000s, local-currency bonds delivered strong returns thanks to rising currencies and high yields. More recently, a stronger U.S. dollar and tighter financial conditions have dampened performance. However, with emerging market currencies now looking undervalued, as the authors suggest, local-currency bonds may once again offer attractive opportunities for investors.
Crypto
One year of Bitcoin spot ETPs: A brief market and fund flow analysis (Oefele)
Since their launch in 2024, Bitcoin spot exchange-traded products have attracted over $100 billion, with most assets concentrated in one product. Investor flows closely follow recent Bitcoin price moves, showing little sensitivity to inflation expectations or stock market trends. This suggests that activity in these products mainly reflects short-term trend-following rather than long-term conviction or macro-driven investment strategies.
Currencies
Mispricing and Risk Premia in Currency Markets (Bartram, Djuranovik, Garratt, and Xu)
Previous research has shown that returns of equity anomalies typically decline by around 50% after publication. This paper conducts a similar analysis of eleven common currency anomalies and finds that most experience a significant drop in returns, about 66%, following academic publication. However, the carry trade anomaly notably stands out with virtually no decline, suggesting that it may primarily represent compensation for risk rather than mispricing.
Equities
Direction is More Important than Speed: A Comparison of Direction and Value Prediction of Stock Excess Returns (Cheng, Shang, and Zhao)
This paper shows that predicting the direction of returns outperforms predicting actual returns across a range of linear and non-linear models. Directional models perform especially well during recessions and in forecasting market downturns. While simple linear models do well, machine learning methods like random forests and gradient boosting often yield better accuracy, though the Sharpe ratio improvements from more complex models are modest. Overall, the results suggest investors can improve performance by focusing on directional signals rather than precise return forecasts.
Market Signals from Social Media (Cookson, Lu, Mullins, and Niessner)
Several papers have found that social media sentiment can predict returns. This study analyzes millions of posts from Twitter, StockTwits, and Seeking Alpha to build daily market-wide indexes that separately capture investor mood (sentiment) and focus (attention), with the latter based on post and article volume. The two signals predict S&P 500 returns in distinct ways: high sentiment precedes short-term reversals, while high attention signals continued declines. A market-timing strategy using these indexes delivers strong risk-adjusted returns.
Momentum at Long Holding Periods (Calluzo, Moneta, and Topaloglu)
Momentum strategies often suffer from high turnover and therefore high trading costs. By leveraging the persistence of momentum rankings and the ability to anticipate future momentum rankings, the authors construct portfolios that outperform the standard approach, even before accounting for trading costs.
Equity Risk Premiums (ERP): Determinants, Estimation, and Implications – The 2025 Edition (Damodaran)
Figuring out how much extra return investors need for taking stock market risk can be tricky. Damodaran reviews three main methods for determining the equity risk premium: Historical returns, investor surveys, and market-based models. He finds that market-implied premiums backed out from current prices, expected cash flows, and default spreads, are the most reliable for forecasting, while historical averages often fail as predictors. For investors, choosing the right estimate can meaningfully impact valuations and decisions.
Short-selling Profitability, Stock Lending Fees, and Asset Pricing Anomalies (Da, Fu, Lin, and Lu)
Some stock market anomalies stick around because it’s costly to bet against overpriced stocks. Short sellers often face steep borrowing fees. This paper finds that shorting stocks most sensitive to short-selling pressure, those whose prices tend to drop the most when short interest rises, is still profitable even after accounting for costs. For investors, this means that anomalies remain exploitable in stocks where the price impact of short sellers is stronger.
Biased Signals? Rethinking Stock Market Indices as Leading Economic Indicators (Hansen)
Major indices like the S&P 500 are widely seen as signals of where the economy is headed. However, their focus on the largest companies might distort the picture. This paper tests whether large-cap indices help distinguish between good and bad future economic conditions and finds they often fail, especially before downturns. In fact, a basket of randomly selected stocks often does better at signaling economic trouble. Investors aiming to anticipate recessions may be better off using broader or more diverse stock samples than relying solely on large-cap indices.
Profitability retrospective: What have we learned? (Medhat and Novy-Marx)
Investors often diversify across strategies like quality, low volatility, and various value signals. Yet many of these approaches overlap. This paper shows that a single factor, profitability, explains the alpha of strategies such as return on equity, earnings stability, low beta, low volatility, and intangibles-adjusted value. For investors, the takeaway is that directly targeting profitability may be more effective and parsimonious than relying on these bundled signals.
Machine Learning and Large Language Models
Bridging Language Models and Financial Analysis (Lopez-Lira, Kwon, Yoon, Sohn, and Choi)
This is an extensive survey paper, exploring how large language models are being applied in finance, offering a broad overview of recent progress. Key areas include predicting stock returns using sentiment analysis, identifying financial risks from disclosures, and generating ESG scores from company reports. It also highlights new frontiers like LLM-based agents that simulate investor behavior.
Assessing Consistency and Reproducibility in the Outputs of Large Language Models: Evidence Across Diverse Finance and Accounting Tasks (Wang and Wang)
Investors and researchers increasingly use AI to analyze financial documents, but these tools can give slightly different results each time, even with the same input. This paper tests AI performance on tasks like judging tone in earnings calls, summarizing MD&As, and predicting future earnings. It finds that simple tasks like classification and sentiment analysis are highly consistent, for instance, models labeled forward-looking statements the same way over 90% of the time. More complex tasks like summarization and earnings prediction showed more variation. Investors can improve reliability by averaging outputs across multiple runs rather than trusting a single result.
From Deep Learning to LLMs: A survey of AI in Quantitative Investment (Cao, Wang, Lin, Wu, Zhang, Ni, and Guo)
Most investing strategies follow a fixed process: clean data, predict returns, build portfolios, and place trades. While these steps are often driven by human-designed models, the paper explains how large language models (LLMs) such as ChatGPT are changing the game. The authors survey the use of autonomous agents capable of reading financial news, extracting signals, and interacting with various tools. Overall, the paper offers a comprehensive discussion of how LLMs can streamline research and improve the design of alpha strategies.
AlphaQuant: LLM-Driven Automated Robust Feature Engineering for Quantitative Finance (Yuksel)
The paper points out that manual feature engineering in finance is often slow, biased, and struggles to adapt to changing markets. It introduces an automated system that uses large language models to generate creative features and refining them through an iterative loop of machine learning evaluation. By repeatedly testing, ranking, and improving features based on predictive performance, the system aims to uncover robust signals while addressing the limitations of manual approaches.
Portfolio Construction
Investment Base Pairs (Goulding and Harvey)
Many investment strategies rank assets by signals like value or momentum and build portfolios by buying and shorting, say, the top and bottom deciles. This paper argues that such approaches overlook how signals interact across assets. By decomposing strategies into all possible two-asset “base pairs,” the authors show that each pair’s return depends on factors like own- and cross-asset predictability, signal correlation, and signal biases. Selecting only the top-ranked pairs based on these features can significantly boost performance.
Volatility
Forecasting U.S. equity market volatility with attention and sentiment to the economy (Halouskova and Lyocsa)
The authors construct sentiment measures from general market and macroeconomic news, derived from news articles and social media posts analyzed using FinBERT. Integrating these sentiment scores into various volatility forecasting models significantly improves forecasting performance for individual stocks, suggesting sentiment is valuable for predicting market volatility.
Blogs
EM sovereign bond allocation with macro risk premium scores (Macrosynergy)
Informational Edge (Quantitativo)
Weekly Research Insights (QuantSeeker)
Crypto ETPs: An Examination of Liquidity and NAV Premium (FEDS Notes)
GitHub
Backtesting.py - “Backtest trading strategies with Python.”
Pandas TA - “Technical Analysis Indicators”
Asset News Sentiment Analyzer - “A sentiment analyzer package for financial assets and securities utilizing GPT models.”
Podcasts
Bob Pardo II - Building Trading Strategies that Work with Walk Forward Analysis (The Algorithmic Advantage)
Nothing Good Happens Below the 200-Day Moving Average ft. Mark Rzepczynski (Top Traders Unplugged)
Unconventional Rules of History’s Best Traders: Practical Lessons for All Investors | Jack Schwager (Excess Returns)
75% of Investors Will Disagree with These Ideas: Evidence Says They are True (Excess Returns)
Last Week’s Most Popular Links
The theory of quantitative trading (Berdondini)
Mispricing and Correction in Short-Term Returns (Han, Kang, and Lee)
Trading Theta: A Strategy Exploiting Time Decay (Lu)
Disclaimer: This newsletter is for informational and educational purposes only and should not be construed as investment advice. The author does not endorse or recommend any specific securities or investments. While information is gathered from sources believed to be reliable, there is no guarantee of its accuracy, completeness, or correctness.
This content does not constitute 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, and these holdings may change at any time 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 should not be considered definitive or comprehensive representations of the original works. Readers are encouraged to refer to the original sources for complete and authoritative information.
This newsletter may contain 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 external sites or for that of any subsequent links. Users access these links at their own risk.
The author assumes no liability for losses or damages arising from the use of this content. By accessing, reading, or using this newsletter, you acknowledge and agree to the terms outlined in this disclaimer.