Paper: arXiv 2602.00073

Authors: Yurui Wu, Qingying Deng, Wonou Chung, Mairui Li

Abstract

Time series encountered in practice are rarely stationary. When the data distribution changes, a forecasting model trained on past observations can lose accuracy. We study a small-footprint test-time adaptation (TTA) framework for causal timeseries forecasting and direction classification. The backbone is frozen, and only normalization affine parameters are updated using recent unlabeled windows. For classification we minimize entropy and enforce temporal consistency; for regression we minimize prediction variance across weak time-preserving augmentations and optionally distill from an EMA teacher. A quadratic drift penalty and an uncertainty triggered fallback keep updates stable. We evaluate this framework in two stages: synthetic regime shifts on ETT benchmarks, and daily equity and FX series (SPY, QQQ, EUR/USD) across pandemic, high-inflation, and recovery regimes. On synthetic gradual drift, normalization-based TTA improves forecasting error, while in financial markets a simple batch-normalization statistics update is a robust default and more aggressive norm-only adaptation can even hurt. Our results provide practical guidance for deploying TTA on non-stationary time series.

Complexity vs Empirical Score

  • Math Complexity: 4.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper provides a rigorous empirical evaluation of Test-Time Adaptation (TTA) frameworks on financial markets and benchmark time-series data. It features solid methodological clarity, statistical testing (Diebold-Mariano), and open-source code, though the underlying mathematical formulations are straightforward application of existing loss functions.

Research Flowchart

  flowchart TD
    A["Research Goal: Robust Test-Time Adaptation (TTA) <br>for Non-Stationary Time Series"] --> B["Data Inputs: ETT Benchmarks (Synthetic Shifts)<br>& Financial Markets (SPY, QQQ, EUR/USD)"]
    
    B --> C["Core Methodology: Freeze Backbone,<br>Adapt Normalization Affine Parameters Only"]
    
    C --> D{"Methodology by Task"}
    
    D -- "Classification" --> E["Minimize Entropy & Enforce<br>Temporal Consistency"]
    D -- "Regression" --> F["Minimize Augmentation Variance,<br>EMA Distillation & Drift Penalty"]
    
    E & F --> G["Key Findings: TTA Improves Synthetic Shifts;<br>Simple Batch-Norm Update Best for Finance"]