Case Study · Data Science
Excel Workbooks Ported to Python
Excel credit models are flexible, but they’re slow, cumbersome, and difficult to maintain and distribute at scale. KPMG earmarked two Excel-based credit models to be rebuilt as something faster, more stable, and fully auditable.
The Challenge
Powerful Models. Fragile Infrastructure.
KPMG’s IFRS9 credit models lived in Excel. Flexible enough for analysts to iterate quickly, but brittle everywhere else.
Distribution meant emailing spreadsheets. Maintenance meant hunting through nested formulas. And every calculation run tied up an analyst’s desktop for the better part of an afternoon.
Two models in particular needed to move: the core IFRS9 credit model, and the IFRS9 hedge effectiveness model used to classify RMB trades. Both needed to keep the auditability regulators expect, while shedding the runtime and maintenance burden Excel had built up over years.
- IFRS9 compliance
- Manual Excel distribution
- Slow calculation runtimes
- Audit trail requirements
- RMB trade classification
Challenge
Solutions Built
Two Models. Off Excel, Onto Python.
Symbyte rebuilt both credit models from the ground up. One as a desktop-ready data science app, the other as a CLI tool wired directly into RMB’s systems.
IFRS9 Credit Model
A Python-backed data science app that enables visual kickoff of credit modelling runs, ready to use on the desktop with no admin access and zero outside installs or dependencies. Calculation run time dropped from 15 minutes to 30 seconds.
IFRS9 Hedge Effectiveness Model
A Python CLI tool integrated directly with RMB’s systems to classify hedge effectiveness trades correctly, cutting run time from 8 hours down to 10 minutes.
30x Faster Credit Modelling
Calculation runtime on the IFRS9 credit model fell from 15 minutes to 30 seconds. A full modelling cycle now takes as long as a coffee refill.
Hedge Trades Classified in Minutes
The IFRS9 hedge effectiveness model's 8-hour run time dropped to 10 minutes, integrated straight into RMB's existing systems.
Zero-Install Desktop Access
Analysts can kick off modelling runs visually, with no admin rights and no outside dependencies to install or manage.
Audit-Ready by Design
Moving off Excel gave KPMG a faster, more stable, and more auditable foundation for both models. Without losing the flexibility analysts relied on.
