Re(Visiting) Time Series Foundation Models in Finance
By: Eghbal Rahimikia, Hao Ni, Weiguan Wang
Potential Business Impact:
Teaches computers to predict stock prices better.
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.
Similar Papers
Time Series Foundation Models for Multivariate Financial Time Series Forecasting
General Finance
Helps predict money changes with less data.
Time Series Foundation Models: Benchmarking Challenges and Requirements
Machine Learning (CS)
Tests if future predictions are truly new.
Evaluating Time Series Foundation Models on Noisy Periodic Time Series
Machine Learning (CS)
AI struggles to predict future patterns with noise.