Model Outputs
From Open Risk Manual
Definition
Model Outputs denotes the list of numerical or categorical variables that encapsulate the outcome of a model calculation or "run". The significance of such outcomes lies in the manner in which they are subsequently used in internal and/or external processes.
Examples
There is a wide variety of possible model outputs:
- Continuous Numerical values (valuations, hedge sizes, continuous risk scores)
- Discrete Numerical values or Categorical variables (ratings)
- Distributions (rankings of possible scenario losses)
- Discrete indicators (yes / no)
- Tabular Data (risk allocation to positions)
Issues and Challenges
- Model outputs are frequently becoming in turn Risk Data (Risk Parameters) that are used as Model Inputs in further models, creating chains of dependencies and possible Model Risk propagation
- Careful documentation of all model outputs in use is part of good Model Governance