Ortec Finance partnered with M&G Investments to pilot its new Scenario-Based Machine Learning (SBML) approach—designed to optimize complex, constraint-driven portfolios within a stochastic modeling framework using the GLASS platform.
Traditional methods struggle with non-linear regulatory constraints, often delivering sub-optimal results. SBML overcomes this by integrating these constraints directly into the optimization process. In this case, it optimized Mean Surplus vs. CVaR 5% Surplus, while constraining Market Risk SCR Charge—uncovering more efficient portfolios than both the current strategy and traditional models.
Key outcomes:
- Higher returns achieved at similar or slightly higher SCR levels
- Non-linear constraints successfully included in optimization
- Fast run time—typically overnight
- Targeted search avoids infeasible solutions
SBML will be available as an additional GLASS module in late 2025, offering insurers and asset managers a flexible, powerful tool for strategic asset allocation.
If you are interested in reading the case study in full, you can download it here.