Multilayer Perceptron Neural Network Models in Asset Pricing: An Empirical Study on Large-Cap US Stocks

ArXiv ID: 2505.01921 “View on arXiv”

Authors: Shanyan Lai

Abstract

In this study, MLP models with dynamic structure are applied to factor models for asset pricing tasks. Concretely, the MLP pyramid model structure was employed on firm-characteristic-sorted portfolio factors for modelling the large-capital US stocks. It was further developed as a practicable factor investing strategy based on the predictions. The main findings in this chapter were evaluated from two angles: model performance and investing performance, which were compared from the periods with and without COVID-19. The empirical results indicated that with the restrictions of the data size, the MLP models no longer perform “deeper, better”, while the proposed MLP models with two and three hidden layers have higher flexibility to model the factors in this case. This study also verified the idea of previous works that MLP models for factor investing have more meaning in the downside risk control than in pursuing the absolute annual returns.

Keywords: Multi-Layer Perceptron, Factor Models, Deep Learning, Asset Pricing, Portfolio Optimization, Equities

Complexity vs Empirical Score

• Math Complexity: 5.0/10
• Empirical Rigor: 7.0/10
• Quadrant: Holy Grail
• Why: The paper applies advanced neural network models (MLP) to financial factor models with clear mathematical formulation, placing it in the moderate-to-high complexity range. It is heavily data-driven, uses specific datasets (large-cap US stocks), evaluates performance across different periods including COVID-19, and proposes a practicable investment strategy, indicating strong empirical rigor.

  flowchart TD
    A["Research Goal: Develop MLP Factor Models<br>for Asset Pricing &amp; Investing Strategy"] --> B["Data: Firm-Characteristic-Sorted Portfolio Factors<br>for Large-Cap US Stocks"]
    B --> C["Modeling: Dynamic MLP Pyramid Structure<br>2 &amp; 3 Hidden Layers"]
    C --> D["Computations: Model Training &amp; Prediction<br>Split by Pre-COVID vs. COVID Periods"]
    D --> E["Investment Strategy: MLP-based Factor Investing"]
    D --> F["Evaluation: Model Performance &amp; Investing Performance"]
    E --> F
    F --> G["Key Findings: Effective Downside Risk Control<br>Performance Gains Over 2-3 Hidden Layers<br>No Benefit from Extremely Deep Models"]