Description

This course introduces a multi-step framework for onchain stablecoin depeg risk monitoring. Publicly indexed data from major onchain trading venues is collected, preprocessed and packaged into a continuously updated dataset. 

A bi-model framework is introduced : 

- The first chapter focuses on an interpretable tree-based early warning classifier. This model output an easy to interpret warning for absolute depeg above a given threshold, and its accompoanying feature attributions for transparency. 
- The second chapter focuses on the implementation of a NN based probabilisitc forecaster. The model derives a full probabilistic forecast of the future stablecoin price, with a focus on tail calibration for VaR/ES evaluation.
- The third chapter focuses on the deployment of both models in a live dashboard, including a multi-agentic LLM framework for result interpretation. Turning informative model outputs into transparent, explainable investor indications.