Foundation Time-Series AI Model for Realized Volatility Forecasting

TL;DR

This paper evaluates the TimesFM foundation model for financial volatility forecasting, demonstrating that incremental fine-tuning significantly enhances its accuracy and outperforms traditional econometric benchmarks in dynamic markets.

Contribution

It introduces a fine-tuning approach for foundation models in volatility forecasting, showing improved performance over pretrained models and standard benchmarks.

Findings

Fine-tuned TimesFM models outperform pretrained versions.

Incremental learning improves volatility forecast accuracy.

Fine-tuned models statistically outperform traditional econometric models.

Abstract

Time series foundation models (FMs) have emerged as a popular paradigm for zero-shot multi-domain forecasting. These models are trained on numerous diverse datasets and claim to be effective forecasters across multiple different time series domains, including financial data. In this study, we evaluate the effectiveness of FMs, specifically the TimesFM model, for volatility forecasting, a core task in financial risk management. We first evaluate TimesFM in its pretrained (zero-shot) form, followed by our custom fine-tuning procedure based on incremental learning, and compare the resulting models against standard econometric benchmarks. While the pretrained model provides a reasonable baseline, our findings show that incremental fine-tuning, which allows the model to adapt to new financial return data over time, is essential for learning volatility patterns effectively. Fine-tuned variants not only improve forecast accuracy but also statistically outperform traditional models, as demonstrated through Diebold-Mariano and Giacomini-White tests. These results highlight the potential of foundation models as scalable and adaptive tools for financial forecasting-capable of delivering strong performance in dynamic market environments when paired with targeted fine-tuning strategies.