FRTB Data Management

The Real Price Observation test (RPO test) checks whether a risk factor has at least 24 actual market prices observed over the preceding 12 months. Only risk factors passing this test are classified as modellable under FRTB and may be calculated using expected shortfall models in the Internal Models Approach (IMA). Non-modellable risk factors (NMRFs) attract a separate, significantly higher capital charge. Systematic market data sourcing that closes RPO gaps is therefore the most effective lever for reducing FRTB capital requirements.

FRTB requires comprehensive market data in three categories: first, historical time series with at least 10 years of data including stress periods for expected shortfall calculation. Second, real-time or end-of-day market data for ongoing risk measurement and P&L attribution. Third, real price observations for the modellability test of each individual risk factor. This includes complete yield curves, volatility surfaces, credit spreads and commodity curves at the granularity level prescribed by regulators.

NMRFs are risk factors that fail the RPO test because too few observable market prices exist. They can account for 30% or more of total FRTB capital requirements, particularly for exotic products and emerging market positions. Reduction strategies include expanding market data sources (e.g. through data pooling initiatives like the DTCC RPO Service), optimising risk factor granularity by grouping related factors, and improving internal data capture for OTC transactions.

The Internal Models Approach (IMA) imposes significantly higher data requirements than the Standardised Approach (SA). IMA needs complete risk factor time series, RPO evidence for each factor, daily P&L data for the P&L attribution test and sufficient data for backtesting. The SA works with predefined sensitivities and primarily requires accurate position data. Many banks run both approaches in parallel, making a unified data foundation essential.

Data governance under FRTB is a regulatory requirement, not an optional framework. Supervisors expect traceable data lineage, defined data ownership, documented quality standards and audit-proof change logs. Every risk factor must be traceable to its data source, data changes must be recorded in an immutable audit trail, and clear escalation processes must exist for data quality issues.

A complete implementation typically takes 12 to 24 months: 3 months for assessment and architecture design, 6 to 12 months for core implementation (data pipelines, governance framework, quality monitoring), and a further 3 to 6 months for fine-tuning, validation and audit preparation. Faster capital relief can be achieved by starting with a targeted NMRF reduction initiative, where initial results are often measurable after 3 to 6 months.

ADVISORI supports the entire process: analysis of your existing data landscape, identification of modellability gaps, assessment of RPO coverage, target architecture design, market data sourcing strategy, data governance framework and automated quality assurance. After go-live, we provide continuous monitoring, regular data quality reviews and preparation for supervisory examinations.