Notes

A running list of articles and papers I’ve found interesting, mostly around cross-sectional asset pricing, machine learning models, and quant investing.

Machine Learning + Asset Pricing

Artificial Intelligence Asset Pricing Models
Kelly et al. (2025)
Uses transformers to learn the stochastic discount factor from raw panel data. Introduces the transformer architecture in a clean way and explicitly learns cross-stock structure.

Building Cross-Sectional Systematic Strategies by Learning to Rank
Poh et al. (2021)
Applies learning-to-rank techniques to directly optimize for cross-sectional ordering instead of predicting returns. Tightly aligned with how quant signals are actually used.

Tabular Foundation Models

TabPFN-2.5: Advancing the State of the Art in Tabular Foundation Models
Grinsztajn et al. (2024)
A pre-trained model for tabular data that learns general patterns across many datasets, then predicts on new tables with little or no tuning. Promising for learning cross-sectional structure, but I’m still worried about scalability and potential information leakage in point-in-time backtests.

TabPFN (YouTube)
Video overview of TabPFN and how it works for tabular prediction.

Return Predictability

How Global is Predictability? The Power of Financial Transfer Learning
Hellum et al. (2023)
Asks whether return predictability is global or country-specific. Key takeaway: predictability is overwhelmingly global (94 to 96% of the signal). Local adjustments help at the margin, but the global model dominates out of sample.

Anomaly Replication

Replicating Anomalies
Hou et al. (2017)
Compiles 447 anomalies and shows most lose significance once microcaps are controlled (NYSE breakpoints, value-weighted returns) and stronger t-cutoffs are applied. Once you remove small and micro-caps, a lot of the “anomalies” just don’t survive.

Risk Management

Risk Everywhere: Modeling and Managing Volatility
Bollerslev et al.
Examines realized volatility patterns across 50+ commodities, currencies, equity indices, and fixed-income instruments. Uses panel-based estimation to achieve superior out-of-sample risk forecasts. Their best vol model only improves performance by about 50 bps versus a simple rolling volatility model.

Limits to Arbitrage / Capacity

Bottom-Up Capacity Constraints and the Limits of Anomaly Profitability
Cartea et al. (2025)
Argues that asset-level capacity constraints sharply limit anomaly scalability. Profitability falls once realistic trading capacity is imposed. One of the few papers that really digs into the mechanics of what’s scalable.