FRTB Backtesting Requirements: Basel III Backtesting Validation and Model Performance Optimization

FRTB backtesting is the daily back-validation of internal market risk models under the Fundamental Review of the Trading Book. It compares predicted Value-at-Risk figures against actual losses incurred over at least 250 trading days. Without passing backtesting, a trading desk cannot apply the Internal Models Approach and must fall back to the standardised approach.

The traffic light approach classifies backtesting results into three zones: the green zone (0‑4 exceptions in 250 trading days) confirms acceptable model quality. The yellow zone (5‑9 exceptions) triggers a supervisory capital surcharge. The red zone (10 or more exceptions) can lead to withdrawal of IMA approval for the affected trading desk. Each exception represents a day on which the actual loss exceeded the predicted VaR.

Under FRTB, backtesting is primarily performed at desk level — each trading desk must individually pass the VaR back-test and the P&L attribution test. A desk that fails falls back to the standardised approach, while other desks can continue using the IMA. Bank-level backtesting assesses overall model quality and feeds into the supervisory capital surcharge.

FRTB backtesting requires daily P&L data at desk level: both hypothetical P&L (based on end-of-day positions with unchanged market data) and actual P&L values. Additionally, daily VaR and Expected Shortfall forecasts, at least 250 trading days of historical data, complete risk factor time series and clear position assignment to trading desks are required.

From 10 exceptions within 250 trading days (red zone of the traffic light approach), the supervisory authority can withdraw IMA approval for the affected trading desk. From 5 exceptions (yellow zone), an additional capital surcharge applies. The red zone generally triggers a mandatory switch to the standardised approach for at least 12 months.

The VaR back-test uses the traffic light approach to check whether the model's loss predictions match actual outcomes — it measures forecast accuracy. The P&L attribution test checks whether the risk factors used in the model sufficiently explain real P&L movements. Both tests must be passed, but they measure different aspects of model quality.

When a trading desk fails backtesting, it must switch from the Internal Models Approach to the standardised approach. This typically results in higher capital requirements since the standardised approach imposes more conservative capital charges. The switch applies for at least 12 months. The bank must submit a remediation plan to the supervisor and can only reapply for IMA after renewed approval.

ADVISORI guides banks through full implementation of FRTB backtesting requirements: gap analysis of existing backtesting processes, building data infrastructure for daily desk-level validations, implementing the traffic light approach and escalation procedures, preparing supervisory documentation and setting up ongoing model quality monitoring.