Weekly Research Recap
Latest research on investing and trading
Here’s this week’s recap, featuring the most actionable investing insights from the past seven days, spanning academic research, industry reports, blogs, and social media, with direct links throughout.
Commodities
Was Allen Paul Right? Liquidation Bias in Commodity Futures Markets (Irwin, Sanders, Smith, and Yan)
As futures contracts approach settlement, the price gap between front- and next-month contracts widens predictably. Across 27 U.S. physically delivered commodity futures (1990–2021), the gap rises about +0.65% in the last 15 trading days, strongest in grains (+0.94%) and livestock (+1.75%), while cash-settled financial futures show no comparable effect. Results are consistent with delivery-related option value collapsing near expiry. Key takeaway: Rolling earlier can reduce exposure to end-of-contract pricing distortions.
Crypto
Beyond Gold: A Fundamental Valuation of Bitcoin as a Non-Sovereign Asset (Krause)
The author values Bitcoin as a cash-flow-free store of value, priced by potential share of a $53.8T pool spanning gold ($30.8T), sovereign reserves ($15.0T), and offshore wealth ($8.0T). A 10-year NPV model with 15%–35% required returns implies $5.3k (bear), $16.0k (base), and $66.5k (bull) per BTC. Today’s price of $90k implies faster adoption and/or lower perceived risk. Key takeaway: BTC is priced on adoption + risk, not scarcity alone.
Rules-Based Investing in the Cryptocurrency Market (Gokcen)
Using crypto data from Messari, the paper finds that size offers the most significant return premium: Small-cap portfolios earn 23% excess return/month versus 11% for large caps, a 12%/month spread. Other common signals, such as momentum, illiquidity, and lottery-type effects, are far less stable. Key takeaway: In this sample, size is the strongest and most reliable crypto anomaly.
Equities
Episodic Factor Pricing (Li, Yuan, and Zhou)
Traditional factor investing assumes factor premia are constant. This paper shows premia arrive in bursts, and proposes a real-time on/off regime indicator based on out-of-sample cross-sectional forecast accuracy. When “on,” the FF3-based forecast long–short strategy earns +1.21%/month (Sharpe 0.95); when “off,” it’s −0.35%/month. Timing boosts Sharpe from 0.50 to 0.88 (FF3) and from 1.18 to 1.56 (9 factors). Key takeaway: Treat factor exposure as conditional, not constant.
How Concentrated is the SP500, really? (Tasitsiomi and Noguer I Alonso)
This paper argues that S&P 500 concentration risk is better measured by return dominance than by index weights. It defines dominance as the R² from regressing SPY returns on an equal-weighted Mag-7 basket, and finds it is currently about 0.83, higher than 99.8% of comparable 7-stock baskets. High-dominance regimes have worse crash risk. Key takeaway: Track return dominance, not index weights, to spot concentration risks in index exposure.
CAPE Ratios and Long-Term Returns (Ma, Marshall, Nguyen, and Visaltanachoti)
CAPE’s weak recent record in forecasting long-term stock returns is largely a measurement problem. By matching today’s index constituents to their own historical earnings and value-weighting firm-level CAPEs, long-horizon predictability strengthens sharply, with 10-year out-of-sample R² rising to about 0.56. Key takeaway: CAPE works much better as a bottom-up, constituent-aligned measure than as a top-down index-level statistic.
Regime-Aware Universal Portfolio (Vlasiuk and Smirnov)
The authors build a bull/bear regime-switching framework: Stay invested in a broad, large-cap stock basket when conditions are supportive, and move to cash when the model signals a deteriorating regime, using daily features from trend, volatility, and macro risk indicators. From 2018 to 2025, this lifts the Sharpe to 1.24 vs. 0.75 for buy-and-hold and materially reduces drawdowns. Key takeaway: Systematic exposure timing can add real value.
Factor MAX and Predictable Factor Returns (Wang and Zeng)
Extreme factor moves predict future factor returns. Sorting equity factors by last month’s maximum daily return (MAX) and going long high-MAX vs low-MAX earns +0.32%/month, and the effect survives controls for 1-month factor momentum. The evidence fits an underreaction story at the factor level. Key takeaway: Use extreme factor days as a simple, implementable timing overlay.
Golden Fair Value: A Market Equilibrium & Correction Indicator (Panchal)
This paper introduces Golden Fair Value (GFV), a 52-week high/low “fair value” anchor scaled by 0.618, designed to flag correction bottoms and market overextension. Using 2008–2025 data, GFV improves regime classification, and a GFV crossover strategy returned +184% versus +137% for moving-average crossovers, with fewer trades and max drawdowns around 19 to 25%. Key takeaway: GFV may offer a simple, rule-based re-entry trigger during selloffs.
Machine Learning and Large Language Models
Enhancing Portfolio Optimization with Deep Learning Insights (Luo and Skufca)
This paper tests whether deep learning can learn portfolio weights directly for a long-only multi-asset portfolio, rather than predicting returns first. Replicating earlier work, the authors achieve roughly a 30% annual return, a Sharpe of around 1.85, and a 11% max drawdown, broadly matching the original results. However, performance becomes less stable in later years and weakens on alternative asset universes. Key takeaway: Generalization across regimes and robustness are the main challenges.
Mutual Funds
A Skew is a Skill: Portfolio Skewness of Mutual Fund Holdings (Drienko, Gao, and Liu)
The authors measure the cross-sectional skewness of the returns of a fund’s underlying stock holdings and find that funds with more right-skewed portfolios earn stronger future performance. In U.S. equity mutual funds (1980–2023), high-skewness funds outperform low-skewness funds by roughly 2.5 to 3% per year after fees. Key takeaway: Holdings-based skewness is a powerful predictor of future fund alpha.
Blogs
Prediction markets and lotteries (and my one simple trick for winning the lottery) (Rob Carver)
Managed Futures ETFs: What the Data Says So Far (QuantSeeker)
The Fallacy of Concentration Risk (Quantpedia)
Shifting Tides in Global Markets: The Reemergence of International Investing (CFA Institute)
How to Avoid Dilution in Microcaps - A Dilution Risk Scorecard (MicroCapClub - Ryan Telford)
Portfolio Optimization (Quantitativo)
Podcasts
How I Built a 1.4-Billion-Dollar Quant Fund - Deepak Gurnani on Founding Versor Investments (Odds on Open)
The World’s #2 Trader | Futures Trading World Champion - Marci Silfrain (Words of Rizdom)
When Signals Matter More Than Stories ft. Nick Baltas (Top Traders Unplugged)
How Investors Fall Into Bias Traps with Economists Richard Thaler & Alex Imas (Masters in Business)
Social Media & Industry Research
Can AI Actually Improve Momentum Strategies? (Alpha Architect)
How To Build a Python Backtester That Runs in Seconds (SystematicLongShort)
Last Week’s Most Popular Links
Tail risk exposure and the cross section of expected stock returns (Nicolas)
A Unified Framework for Anomalies based on Daily Returns (Cakici, Fieberg, Neszveda, Bianchi, and Zaremba)
Extracting Forward Equity Return Expectations Using Derivatives (Clark, Lu, and Tian)
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.
On the deep learning portfolio optimization paper (Luo & Skufca) - thanks for flagging it. It inspired me to pull the full PDF. I think it is a Georgia Tech student project that stress-tested an Oxford paper, and it's exactly how research should work. The headline numbers were replicated, but they didn't stop there: extend through COVID or swap asset classes and the edge disappears. Different assets actually underperformed basic MVO. They were honest about the fragility - that's the real finding. More papers should do this.