Market Impact Estimation via Causal Foundation Models

Name: Market Impact Estimation via Causal Foundation Models
Uploaded: 2026-10-19T06:00:17.683981+00:00
Duration: 550 s
Description: The research developed a Causal Prior-Data Fitted Network (PFN) to estimate market impact by modeling hidden informational flow that confounds standard models. This model tracks true parameters at high confounding levels and shows zero-shot transfer to real markets, demonstrating causal inference in trading without relying solely on traditional training methods.

The research developed a Causal Prior-Data Fitted Network (PFN) to estimate market impact by modeling hidden informational flow that confounds standard models. This model tracks true parameters at high confounding levels and shows zero-shot transfer to real markets, demonstrating causal inference in trading without relying solely on traditional training methods.

Open the full episode