Weekly Research Recap
Latest research on investing and trading
Hi there, it’s been another week packed with great papers. Here’s your weekly research recap, with direct links to the full articles.
Commodities
Is Social Media Information Noise or Fundamentals? Evidence from the Crude Oil Market (Ma, Tourani-Rad, Xu, and Zhou)
Social media sentiment from Thomson Reuters MarketPsych Indices predicts crude oil returns, with a one-standard-deviation rise implying a next-day gain of roughly 21 bps. Positive sentiment reflects fundamentals, persisting for months and forecasting inventory changes, while negative sentiment is noise-driven and reverses. A sentiment-based timing strategy delivers a 48% annual return and a 0.93 Sharpe ratio (pre costs).
Crypto
Trading Games: Beating Passive Strategies in the Bullish Crypto Market (Palazzi)
The author presents a cointegration‐based pairs trading strategy on ten major cryptocurrencies (2019–2024) yielding a 71% annual return, Sharpe of about 2, and 14% max drawdown, outperforming buy‐and‐hold and standard pairs trading. Dynamic lookback optimization, volatility filters, and trailing stops sharply reduce drawdowns, enabling the strategy to remain profitable across both bull and bear markets.
Psychological Anchoring Effect and Cross Section of Cryptocurrency Returns (Jia, Simkins, Yan, Zhang, and Zhao)
Cryptocurrencies near their 52-week highs earn substantially higher short-term returns, consistent with investor anchoring bias rather than risk or frictions. A long–short “Nearness52” strategy delivers a 1.4% value-weighted weekly return and stays profitable after costs. Adding the cANCHOR factor to standard crypto models enhances Sharpe ratios and improves anomaly pricing.
Strategic Cryptocurrency Disclosures and Insider Trading (Huang and Gao)
The paper finds that managers of non-crypto firms sometimes add speculative cryptocurrency language to 10-K filings to exploit investors’ enthusiasm for crypto and profit through insider trading. Using ChatGPT-4o to identify such disclosures (2013–2023), the authors show insiders are significantly more likely to buy shares beforehand, executing larger, more opportunistic, and more profitable trades, effects that intensify during Bitcoin market rallies.
Equity
Market-Based Short-Rate Uncertainty and Time-Varying Expected Returns (Yu, Huang, and Yin)
The cyclical component of market-based short-rate uncertainty (SRU), derived from Eurodollar options, strongly predicts equity returns, while its slow-moving trend offers no value. Out-of-sample, cyclical SRU delivers Sharpe ratios of 0.76–1.10, outperforming standard predictors. Its predictive power stems primarily from forecasting future cash flows, making short-rate uncertainty shocks a potential signal for market timing and asset allocation.
The Term Structure of Expected Market Exposure (Gao)
Using option‐implied volatilities, the paper estimates each stock’s short‐ (30‐day) and long‐term (6‐month) exposure to market volatility by regressing its implied variance on the market’s. The signal is the difference between these exposures. Stocks with greater short‐term relative to long‐term exposure are riskier and earn higher returns, with high–low spreads up to 15% annually, especially when market volatility and risk aversion are high.
ESG
The Shades of Investment Factors (Sauzet and Zhu)
The paper finds that many common investment factors exhibit strong, unintended ESG biases where, for example, size, value, accruals, and seasonality tilt “brown,” while profitability, low risk, and quality tilt “green.” Excluding the worst ESG performers significantly improves sustainability profiles with minimal performance impact. Hence, the findings suggest investors can align factor strategies with ESG goals without materially sacrificing returns.
Machine Learning and Large Language Models
Unraveling Asset Pricing with AI: A Systematic Literature Review (Chen, Zhang, Xie, Zhang, and Li)
The paper systematically reviews 782 studies on applying AI to asset pricing, identifying three main factor categories: Firm fundamentals/macro indicators, media sentiment, and related‐firm linkages. It finds that deep learning models, especially LSTM, CNN, and GNN, often outperform both classical ML and econometric models in prediction accuracy. Moreover, the authors launch FactorWiki, a database of 1,231 real‐time factors with API access, to standardize testing.
The Market's Mirror: Revealing Investor Disagreement with LLMs (Bhagwat, Cookson, Dim, and Niessner)
Using 216 LLM-simulated investor profiles, the paper measures disagreement as the dispersion of their sentiment scores (1–10) on 200k+ S&P 500 news headlines. Disagreement peaks around political, social, and economic shocks, predicts higher same-day trading volume, and supports a long–short strategy sorted on three-month trailing disagreement, earning about 0.20% monthly alpha.
Is Complexity Virtuous? (Elmore and Strauss)
The paper disputes claims that extreme model complexity improves investment performance. Replicating Kelly et al. (2024) with 12,000 random Fourier features, the authors find that heavy shrinkage in complex ridge regressions drives forecasts toward simple moving averages of past returns, erasing any predictive signal. Sharpe ratios of complex models remain below buy-and-hold and frequently trail those from simpler models.
Macro
Stagflationary Stock Returns (Knox and Timmer)
Earlier research has shown that inflation shocks tend to lower equity prices. This paper confirms that result and shows investors interpret such shocks as supply‐driven cost increases, consistent with a stagflationary view, which reduce real cash flows and raise equity risk premia. The effect is smaller for firms with high market power. A one percentage point positive inflation surprise lowers the stock market by about 2.75% in nominal terms over five days.
Mutual Funds
Tactical Allocation for Vanguard Investors: A Defensive Strategy for Retirement Portfolios (Carlson)
From 2000 to 2025, the paper presents a rules-based dual momentum strategy rotating among four Vanguard funds covering growth, income, inflation protection, and market-neutral exposures. Each fund receives a 25% weight if its average 1-, 3-, 6-, and 12-month return exceeds the 90-day T-bill; otherwise, that weight moves to T-bills. The strategy delivered a 6.16% CAGR with 4.09% volatility, Sharpe 1.02, and minimal crisis drawdowns. Takeaway: Simple trend filters across diverse assets can achieve equity-like returns with bond-like risk.
Options
In Search of Seasonality in Intraday and Overnight Option Returns (Bali, Goyal, Moerke, and Weigert)
The paper finds a clear intraday–overnight cycle in delta-hedged option returns: Same-period momentum and cross-period reversal yield 0.22%–0.45% per half-day (pre-costs), persisting up to 20 days and strengthening over time. For example, buying at-the-money options with the top five prior intraday returns and shorting the worst earns 0.23% intraday. The effect stems from implied-volatility shifts and peaks when market makers face capacity constraints.
Portfolio Construction
Return Stacking and Portable Alpha, an Investor's Guide (Basnicki and Pickering)
Return stacking overlays uncorrelated strategies, like CTAs, on top of core equity or bond exposure, enhancing diversification without reducing market participation. Long-term examples such as PIMCO StocksPLUS and Millburn Catalyst show outperformance. For example, PIMCO’s PSLDX earned 12.38% annually since 2007 vs. 10.34% for the S&P 500 TR, with similar drawdowns. Effective use requires avoiding equity overconcentration, monitoring conditional correlations, and prioritizing divergent strategies that perform in crises.
Volatility
Using S&P 500 high-frequency volatility data from 1996 to 2023, the study finds that smooth-transition and threshold HAR models consistently outperform both linear benchmarks and advanced machine learning across multiple performance measures, particularly with limited predictors and during market stress. Overall, the results suggest that simple, regime-switching econometric models can outperform complex machine learning in practical volatility forecasting.
Do Recession Fears Help to Predict Stock Market Volatility? International Evidence (Ma, Zhang, Zhou, and Zhou)
A Google‐search–based Recession Fear Index (RFI) forecasts next‐month stock volatility across 11 developed markets, boosting out‐of‐sample R-squared by up to 6.3% (3.3% global average). The signal persists after macro and uncertainty controls, driven mainly by U.S. recession fears and finance‐related searches. Tracking such spikes can provide an early volatility warning.
Blogs
Overnight Returns: Risk or Conspiracy? (Eric Falkenstein)
Skewness as a Commodity Signal (QuantSeeker)
Cultural Calendars and the Gold Drift: Are Holidays Moving GLD ETF? (Quantpedia)
GitHub
algo-trader - Trading bot
qf-lib - Backtesting and related tools
investing-algorithm-framework - Backtesting engine
Podcasts
Beyond Traditional Trend: Leveraging Experience, Short Term, and Crypto with Mike Stendler (RCM Alternatives)
Mark Anson, CFA: Rethinking Portfolio Construction in a Globalized, Uncertain World (CFA Institute)
You're Copying the Wrong Buffett | Kai Wu on His Real Edge (Excess Returns)
Social Media
There Ain’t No Such Thing as a Free Lunch (Cliff Asness, AQR)
Hindsight and Survivorship Biases in Managed Futures (Larry Swedroe)
Last Week’s Most Popular Links
Price Pressures from Daily Mutual Fund Flows: Does it pay to go with the flow? (Suominen and Tuovinen)
The case for low-risk equity investing: evidence from 2011-2025 (Leote de Carvalho, Andreis, and Laplenie)
A tug of war across the market: overnight-vs-daytime lead-lag networks and clustering-based portfolio strategies (Lu, Zhang, Reinert, Cucuringu)
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.