Description
For a major bank in Singapore, the rollout of a Monte Carlo–based XVA and PFE framework required a rigorous independent model validation covering the entire computation chain.
This represented a significant undertaking, with trade pricing being the most challenging area, due to the need to generate and validate market value grids for a wide range of products spanning all major asset classes.
A dedicated and reliable approach was required to ensure the model validation is completed thoroughly without jeopardizing the overall project timelines.
Key challenges
- Very broad coverage of asset classes (rates, FX, commodities, credit, equity, security financing trades) and products, including complex exotics, making trade pricing validation the main model validation bottleneck;
- Tight timelines aligned with the XVA and PFE program roadmap.
Alef's approach: Turnkey model validation library
Alef designed and delivered a fully documented Python pricing library, to be used as a challenger library for trade pricing validation by the bank's model validation team.
It was designed to be:
- Clear and transparent, usable immediately, and easy to review by the audit teams and regulator.
- Easily extensible, to include new products and asset classes in the future.
The delivery was complemented by:
- A robust testing framework safeguarding against regressions during extensions by the model validation team.
- Methodology documentation detailing all the implemented models and assumptions.
- Workshops covering the theoretical details, code architecture and implementation.
- Ongoing support.
This ensured both immediate delivery success and long-term reuse of the library.
Furthermore, all findings uncovered during the development were raised in a timely manner with very precise test cases that required minimal analysis and allowed a swift resolution.
Products scope
- Closed-form evaluation of:
- IBOR and RFR swaps, including Xccy Swap and Xccy Swap with FX Reset;
- IBOR and RFR caps/floors and European swaptions;
- FX outrights and NDFs;
- FX vanilla and digital options;
- FX barrier and touch options, including single and double barriers;
- Equity vanilla and digital options;
- Commodity swaps and forwards, including futures-based commodities and precious metals;
- Credit default swaps;
- Bond forwards;
- Bond repos.
- American Monte Carlo (AMC) evaluation:
- Bermuda swaptions and callable swaps;
- FX TARFs;
- Equity autocalls;
- Commodity TARNs.
Technical Stack
- Core Development: Python (NumPy, Pandas, SciPy, lxml, etc.).
- Interactive Analysis: JupyterLab, Plotly.
Key benefits
- Model validation completion on schedule, with no impact on the overall XVA and PFE programs delivery.
- A scalable and transparent validation framework applicable across multiple asset classes.
Key takeaways
- Alef’s deep expertise in XVA and PFE modelling, combined with strong Python capabilities, enabled delivery in record time and at a competitive cost, despite a broad product scope including complex exotics.
- The delivered library remains a long-term, reusable asset for the model validation team.