The Fundamental Review of the Trading Book (FRTB) is a complete overhaul of the Market Risk capital framework, developed by the Basel Committee on Banking Supervision (BCBS).
It was initially ‘finalised’ in January 2016 as part of BCBS 352, but was subsequently refined through further later updates (e.g. BCBS 457 in 2019).
FRTB aims to address the shortcomings of the Basel 2.5 framework and enhance the credibility, consistency, and comparability of Market Risk capital across all banks globally.
Why Do We Need FRTB?
The requirement for banking institutions to implement FRTB came as a response to vulnerabilities of the existing Market Risk framework exposed during the 2007-08 Financial Crisis, where capital held by banks was insufficient to cover their losses.
The key issues with Basel 2.5 were:
- Excessive variability in capital due to inconsistent internal models across institutions.
- An overly simplistic and non-risk-sensitive Standardised Approach.
- Capital arbitrage through lenient transfers between the trading and banking books.
The purpose of FRTB was to correct these flaws, by introducing more risk-sensitive methodologies and stringent governance standards.
When Does FRTB Need to be Implemented?
While the original implementation date was the 1st January 2022, many jurisdictions have opted for phased rollouts. As of 2025, several major economies have begun or are finalising local adoption (e.g. EU with RR3/CRD6, US with NPR all expected soon), with full implementation across large international banks expected within the next 12-18 months (i.e. by end of 2026).
Two Pillars: Internal Model Approach (IMA) vs Standardised Approach (SA)
FRTB mandates capital be calculated under either:
- Internal Model Approach (IMA): Conditional on regulatory approval at the trading desk level.
- Standardised Approach (SA): The default fallback and mandatory for all banks, even those approved to use IMA.
Internal Model Approach (IMA)
Under the Internal Model Approach, the primary metrics, and underlying assumptions will change:
| Feature | Basel 2.5 | FRTB IMA |
|---|---|---|
| Risk Metric | VaR & Stressed VaR | Expected Shortfall (ES) |
| Holding Period | 1 or 10-day Holding Period | Product-specific Liquidity Horizons |
| Diversification | Full Diversification | Constraints on diversification (NMRFs) |
| Credit Risk Treatment | IRC includes migration and default | Default Risk Charge (DRC): Jump-to-Default only |
| Model Approval | Model approval at bank level | Model approval at desk level |
| Validation | Backtesting for eligibility | Backtesting + P&L Attribution (PLA) |
Expected Shortfall
When it comes to Market Risk, Value-at-Risk (VaR) has long been the go-to metric, but VaR has a blind spot. It tells you how bad things might get at a certain confidence level (say 99%), but not how much worse things can get beyond that threshold. That’s where Expected Shortfall (ES) steps in.
Expected Shortfall (also called Conditional VaR) measures the average loss in the worst-case scenarios – the losses beyond the VaR point. In effect, it is trying to answer the question: “If we end up in the worst 2.5% of outcomes, how much do we typically lose?”
Under the FRTB rules, Expected Shortfall replaces VaR as the regulatory standard for the Internal Models Approach, using a 97.5% confidence level, and represents the average of all losses beyond the point that is taken as the current VaR.
Stressed Expected Shortfall & Liquidity Horizons
Risk factors are grouped and assigned liquidity horizons to reflect time required to exit positions during market stress.
| Risk Factor | Liquidity Horizon (Days) |
|---|---|
| Large Cap Equity | 10 |
| Small Cap Equity | 20 |
| Equity Volatility (Large Cap) | 20 |
| Credit – Investment Grade | 60 |
| Credit – High Yield | 120 |
| Credit – Structured (e.g. CDOs) | 250 |
| Interest Rate (vanilla) | 20 |
| Interest Rate Volatility | 60 |
| FX & FX Volatility | 60 |
| Commodities | 20–120 |
The Stressed Expected Shortfall (SES) must be calculated across these horizons using the following formula:
Non-Modellable Risk Factors (NMRFs)
There is a clear distinction between modellable and non-modellable risk factors as part of FRTB.
A Non-Modellable Risk Factor (NMRF) is one that lacks sufficient, reliable market data. Specifically, it must have at least 24 real price observations over the past year, with no gap longer than one month. Only executed trades or firm committed quotes count, and indicative prices are excluded.
As these risk factors can’t be modeled with confidence, they’re excluded from the Expected Shortfall (ES) calculation under the Internal Models Approach, and as a result, banks are required to hold capital add-ons against them, based on stress scenario simulations and conservative assumptions, that often lead to materially higher capital charges.
Regulators also require a complete audit trail, including historical pricing data and sources, going all the way back to 2005, to support modellability assessments.
Ultimately, the better quality your data is (in terms of accuracy and completeness), the lower your capital cost. FRTB makes data integrity not just a technical detail, but a capital decision.
Default Risk Charge (DRC)
Under FRTB, the Default Risk Charge (DRC) replaces the Incremental Risk Charge (IRC), but with a narrower, more focused scope. DRC is designed to purely capture Jump-to-Default (JTD) risk which is the sudden, total loss that occurs if a counterparty or issuer defaults unexpectedly.
Unlike Expected Shortfall, which looks at broader market risk, DRC deals purely with default events, using a 99.9% confidence level over a 1-year time horizon, making it a highly conservative capital measure.
It’s built on a credit risk engine and requires:
- 10 years of historical data to model default correlations across issuers, and
- Calibration aligned with the Internal Ratings-Based (IRB) approach, including Probability of Default (PD) and Loss Given Default (LGD) assumptions.
DRC applies to credit-sensitive positions across trading desks such as bonds, CDS, securitisations, and is a key component of the Internal Models Approach (IMA) under FRTB.
Trading Desk-Level Model Approval
1. P&L Attribution Test (PLA)
The P&L Attribution Test (PLA) is a critical assessment in the FRTB Internal Models Approach, to ensure that a trading desk’s risk models genuinely reflect how the desk makes and loses money.
It does this by comparing two types of daily profit and loss:
- Hypothetical P&L (HPL): Generated by the front office using actual trades and market moves.
- Risk-Theoretical P&L (RTPL): Produced by the risk models, using risk factors included in the capital framework.
If these two diverge too often (i.e. more than 4 exceptions over a 12-month window), then the desk fails the PLA test.
In the case that the PLA test is failed, then the desk must switch to the Standardised Approach which often results in significantly higher capital requirements.
Ultimately, the PLA test ensures that internal models are not just theoretically sound, but tied to the desk’s real-world risk and returns.
2. VaR Backtesting
VaR Backtesting is an important part of market risk validation already, which a daily check on whether a desk’s Value-at-Risk (VaR) model is performing as expected.
The test compares the desk’s Hypothetical or Actual P&L against its 1-day VaR estimate. If the model is working well, the number of breaches, where losses exceed the predicted VaR, should be between 2 and 3 for a 1-day, 99% confidence level.
For FRTB, the backtesting check is looking to see whether the VaR model has more than 12 breaches at 99% confidence, or more than 30 breaches at 97.5% confidence. In the case that either of these are true, this indicates model failure.
Rather than disqualifying the model entirely, a failed VaR backtest leads to a capital penalty, where the VaR multiplier increases, inflating capital requirements to reflect model inaccuracy.
In essence, VaR backtesting is about accountability, to prove that your model can hold up under real-world stress.
Standardised Approach (SA)
Even if a desk passes PLA/VaR tests, the Standardised Approach is still required for reporting.
Sensitivities-Based Method (SBM)
The Sensitivity-Based Method (SBM) is the primary contributor behind capital calculations under FRTB’s Standardised Approach. It measures how sensitive a portfolio is to movements in market risk factors, and translates those sensitivities into regulatory capital.
The SBM breaks risk down into three components:
- Delta (linear risk),
- Vega (volatility risk), and
- Gamma (non-linear or curvature risk).
These are computed for each risk class individually (i.e. we will have Equity Delta, Vega and Gamma calculated, as with other asset classes).
To account for market uncertainty, capital is calculated under three correlation scenarios: low, medium, and high, where the correlation with the maximum result being taken as the final charge
One key feature of the SBM is that no diversification is allowed across different risk classes. This keeps the capital requirement high when portfolios span multiple asset types, reflecting potential contagion in times of stress.
Default Risk Charge (DRC-SA)
The Default Risk Charge (DRC) under FRTB’s Standardised Approach takes a conservative view of the Jump-to-Default (JTD) risk, which as explained above, is the sudden loss from an issuer or counterparty defaulting.
While similar in spirit to the IMA version, DRC-SA does not rely on complex internal models. Instead, it simulates JTD impacts across three product types:
- Correlation Trading Portfolios (CTP),
- Non-CTP securitisations, and
- All other credit-sensitive products.
Each group is capitalised separately using a bucketed aggregation approach. Importantly, there’s no diversification allowed across buckets, ensuring the result remains conservative and robust.
While the calculation of the DRC-SA is relatively simple, it often results in a higher capital outcome compared to the IMA calculation.
Residual Risk Add-On (RRAO)
The Residual Risk Add-On (RRAO) under FRTB is designed to capture the risks that models can’t easily quantify, such as rare, exotic, or structurally unhedgeable exposures.
It applies to trades with complex features such as:
- Gap risk (sudden price jumps),
- Correlation risk (e.g. in bespoke CDO tranches),
- Behavioural or optionality risk (like mortgage prepayment),
- Natural catastrophe triggers (in insurance-linked securities).
Rather than using sensitivities or scenarios, RRAO is calculated simply as Flat Notional x Risk Weight.
This ensures a consistent buffer for tail events and unconventional products where uncertainty is too great for traditional modelling.
Data & Infrastructure Impact
While FRTB reshapes how banks measure risk, it also demands a major leap in data and infrastructure behind the scenes.
To comply, banks face a 3x increase in data processing and storage, driven by granular risk factor modelling, desk-level capital attribution, and historical testing.
FRTB requires:
- Auditable time series dating back to 2005,
- Clear separation of risk factors, desk-level P&Ls, and liquidity horizons, and
- Infrastructure that can support high-frequency, high-volume calculations across multiple approaches.
Trading Book vs Banking Book Transfers
FRTB tightens the rules around transfers between the Trading Book and Banking Book to close the door on capital arbitrage.
Previously, firms could shift risk between books to reduce capital, especially by using internal hedges, however, under FRTB, this is heavily restricted:
- Transfers must be well-documented and justifiable,
- Internal hedges only provide capital relief if they are independently verified,
- In most cases, no automatic capital benefit is granted for offsets between books.
The aim is to have clearer separation, more consistent capital, and far less room for engineering gains behind the scenes.
2025 Outlook
FRTB is no longer a concern in the far future as in 2025 marks the transition from planning to full-scale execution.
Regions such as the EU, Switzerland, Singapore, Hong Kong, and parts of the Middle East are actively pushing toward formal adoption of FRTB, with final rules either in effect or imminent. The Basel Committee’s vision is finally becoming operational reality.
For banks, this means:
- Finalising model approvals and desk-level eligibility for IMA,
- Completing system builds for risk factor eligibility testing (RFET), P&L attribution (PLA), and capital reporting,
- Fully aligning data lineage, governance, and documentation for regulatory inspection.
By 2026, full compliance is expected. While the burden is high, there is a significant strategic upside that firms can get if they get this right, to unlock capital efficiency, model flexibility, and regulatory credibility.
FRTB is the most comprehensive overhaul of market risk rules in decades, not just a tweak to VaR, but a wholesale redesign of how risk is measured, attributed, and capitalised.