Interventional Time Series Priors for Causal Foundation Models

Dennis Thumm, Ying Chen

Keywords: causal inference, time series, structural causal models, prior-data fitted networks, interventional data, synthetic data generation, regime-switching dynamics, foundation models

TL;DR: We propose CausalTimePrior, a framework for generating synthetic temporal SCMs with paired observational and interventional time series, enabling training of causal foundation models for time series.

Abstract: Prior-data fitted networks (PFNs) have emerged as powerful foundation models for tabular causal inference, yet their extension to time series remains limited by the absence of synthetic data generators that provide interventional targets. Existing time series benchmarks generate observational data with ground-truth causal graphs but lack the interventional data required for training causal foundation models. To address this, we propose \textbf{CausalTimePrior}, a principled framework for generating synthetic temporal structural causal models (TSCMs) with paired observational and interventional time series. Our prior supports configurable causal graph structures, nonlinear autoregressive mechanisms, regime-switching dynamics, and multiple intervention types (hard, soft, time-varying). We demonstrate that PFNs trained on CausalTimePrior can perform in-context causal effect estimation on held-out TSCMs, establishing a pathway toward foundation models for time series causal inference.

Track: Research Track (max 4 pages)

Submission Number: 106