Research on statistical learning in finance, including model objectives, overfitting, and portfolio use.
Curated by Signal & Evidence. Read our research approach for how we distinguish author findings from house tests.
The Axiomatic Trader connects model complexity, research search and position sizing through explicit assumptions.
Study: The Axiomatic Trader: Latent Regularity, Information Budgets, and the Canonical Form of a Quantitative Investment System
Feature structure reduces one model’s parameter burden, but the stronger portfolio claim depends on weighting and costs.
Study: Dependence-Informed Sparse Neural Architecture for Stock Return Prediction
MINGLE builds a graph from shared exposures. The gains are promising, but crisis protection remains limited.
Study: Beyond Co-Movement: Locality by Exposures Enables a Joint Factor-Graph Framework for Portfolio Diversification
A compact neural risk model improves the authors’ backtest. Its drawdowns and universe rules deserve equal billing.
Study: Neural Network-Driven Volatility Drag Mitigation under Aggressive Leverage
A neural covariance estimator reports lower risk, but the implementation deserves an equally careful audit.
Study: End-to-End Neural Shrinkage of Indefinite Pairwise Correlation Matrices for Small-Cap-Inclusive Portfolios
A direct route from features to portfolio weights exposes the cost of estimation error.
Study: KellyBoost: Growth-Optimal Portfolio Construction with Gradient-Boosted Trees
We changed one training choice and kept the rest of the experiment fixed.
Study: Getting the Target Right in Return Prediction
