Feature Engineering Methods on Multivariate Time-Series Data for Financial Data Science Competitions

Paper: arXiv 2303.16117

Authors: Thomas Wong, Mauricio Barahona

Abstract

This paper is a work in progress. We are looking for collaborators to provide us financial datasets in Equity/Futures market to conduct more bench-marking studies. The authors have papers employing similar methods applied on the Numerai dataset, which is freely available but obfuscated. We apply different feature engineering methods for time-series to US market price data. The predictive power of models are tested against Numerai-Signals targets.

Complexity vs Empirical Score

• Math Complexity: 3.0/10
• Empirical Rigor: 3.5/10
• Quadrant: Philosophers — conceptual discussion, limited math and data

Why this score: This submission is an incomplete work-in-progress seeking collaborators and datasets rather than presenting a finished empirical study. It lacks the mathematical depth, detailed methodology, and comprehensive benchmarking required for a higher assessment.

Research Flowchart

  flowchart TD
    A[Research Goal: Benchmark Feature Engineering for Financial Time-Series] --> B[Input Data: US Equities/Futures Prices & Numerai-Signals Targets]
    B --> C[Apply Time-Series Feature Engineering Methods]
    C --> D[Predictive Modeling & Computational Benchmarking]
    D --> E[Outcome: Models Evaluated on Numerai Targets]
    E --> F[Call for Collaborators: Seeking Unobfuscated Datasets]