Online learning techniques for prediction of temporal tabular datasets with regime changes

Paper: arXiv 2301.00790

Authors: Thomas Wong, Mauricio Barahona

Abstract

The application of deep learning to non-stationary temporal datasets can lead to overfitted models that underperform under regime changes. In this work, we propose a modular machine learning pipeline for ranking predictions on temporal panel datasets which is robust under regime changes. The modularity of the pipeline allows the use of different models, including Gradient Boosting Decision Trees (GBDTs) and Neural Networks, with and without feature engineering. We evaluate our framework on financial data for stock portfolio prediction, and find that GBDT models with dropout display high performance, robustness and generalisability with reduced complexity and computational cost. We then demonstrate how online learning techniques, which require no retraining of models, can be used post-prediction to enhance the results. First, we show that dynamic feature projection improves robustness by reducing drawdown in regime changes. Second, we demonstrate that dynamical model ensembling based on selection of models with good recent performance leads to improved Sharpe and Calmar ratios of out-of-sample predictions. We also evaluate the robustness of our pipeline across different data splits and random seeds with good reproducibility.

Complexity vs Empirical Score

• Math Complexity: 4.5/10
• Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
• Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper provides a practical and empirical framework for portfolio ranking using online learning and dynamic ensembling to handle market regime changes. While mathematically light, it offers strong empirical rigor with robustness checks across random seeds and temporal data splits.

Research Flowchart

  flowchart TD
    A[Research Goal: Robust Ranking & Prediction on Non-Stationary Temporal Tabular Data] --> B[Input Data: Financial Stock Portfolio Panel Data]
    B --> C[Modular Pipeline: Train GBDTs with Dropout & Neural Networks]
    C --> D[Online Learning Techniques Post-Prediction]
    D --> E[1. Dynamic Feature Projection]
    D --> F[2. Dynamical Model Ensembling based on Recent Performance]
    E & F --> G[Key Outcomes: Reduced Drawdown, Improved Sharpe/Calmar Ratios, High Reproducibility]