Time-Series Foundation AI Model for Value-at-Risk Forecasting

TL;DR

This paper evaluates a pre-trained time-series foundation AI model for Value-at-Risk forecasting, demonstrating that fine-tuning enhances accuracy and outperforms traditional econometric models in risk prediction tasks.

Contribution

It is the first to analyze a foundation AI model for VaR forecasting and shows fine-tuning significantly improves its performance over zero-shot application.

Findings

Fine-tuned foundation model outperforms traditional models in backtesting.

Model performs comparably to the best econometric models in quantile scoring.

Foundation model ranks as top or among top performers across multiple quantiles.

Abstract

This study is the first to analyze the performance of a time-series foundation AI model for Value-at-Risk (VaR), which essentially forecasts the left-tail quantiles of returns. Foundation models, pre-trained on diverse datasets, can be applied in a zero-shot setting with minimal data or further improved through finetuning. We compare Google's TimesFM model to conventional parametric and non-parametric models, including GARCH and Generalized Autoregressive Score (GAS), using 19 years of daily returns from the SP 100 index and its constituents. Backtesting with over 8.5 years of out-of-sample data shows that the fine-tuned foundation model consistently outperforms traditional methods in actual-over-expected ratios. For the quantile score loss function, it performs comparably to the best econometric model, GAS. Overall, the foundation model ranks as the best or among the top performers across the 0.01, 0.025, 0.05, and 0.1 quantile forecasting. Fine-tuning significantly improves accuracy, showing that zero-shot use is not optimal for VaR.