金融时序模型预训练效果不佳，自建金融数据训练才有效。
Re(Visiting) Time Series Foundation Models in Finance
- 用金融数据从头预训练时序模型，提升预测能力。
- 在真实市场数据上，自训练模型显著优于现成模型。
- 数据量、合成数据和调参能进一步提升性能，适合量化研究者。
金融时序预测对交易、组合优化和风险管理至关重要，但受噪声大、非平稳性和异构性影响，仍具挑战。近期受大语言模型启发的时序基础模型（TSFMs）为从大规模多样化数据中学习通用时序表征提供了新范式。本文首次对全球金融市场中的TSFMs进行了全面实证研究。基于涵盖多个市场的日度超额收益大规模数据集，评估了零样本推理、微调及从头预训练相对于强基准模型的表现。结果发现，直接使用的预训练TSFMs在零样本和微调设置下表现较差；而仅在金融数据上从头预训练的模型则实现显著的预测与经济收益提升，凸显领域特定适配的重要性。增加数据规模、引入合成数据增强及超参数调优可进一步改善性能。
原文摘要 · Abstract (English)
Financial time series forecasting is central to trading, portfolio optimization, and risk management, yet it remains challenging due to noisy, non-stationary, and heterogeneous data. Recent advances in time series foundation models (TSFMs), inspired by large language models, offer a new paradigm for learning generalizable temporal representations from large and diverse datasets. This paper presents the first comprehensive empirical study of TSFMs in global financial markets. Using a large-scale dataset of daily excess returns across diverse markets, we evaluate zero-shot inference, fine-tuning, and pre-training from scratch against strong benchmark models. We find that off-the-shelf pre-trained TSFMs perform poorly in zero-shot and fine-tuning settings, whereas models pre-trained from scratch on financial data achieve substantial forecasting and economic improvements, underscoring the value of domain-specific adaptation. Increasing the dataset size, incorporating synthetic data augmentation, and applying hyperparameter tuning further enhance performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。