SA-CVA Under FRTB: Standardized CVA Capital with Claude AI (2026)

SA-CVA and the FRTB-CVA Framework

The Basel IV FRTB-CVA (Fundamental Review of the Trading Book — Credit Valuation Adjustment) framework replaced the old Basel III CVA capital charge with a more risk-sensitive regime effective across most major jurisdictions by 2025. The framework has two approaches: BA-CVA (Basic CVA), a simplified charge for banks without full CVA sensitivity infrastructure, and SA-CVA (Standardized CVA), a sensitivity-based approach for banks with active CVA desks. SA-CVA is substantially cheaper for banks that hedge CVA, but it requires computing a full set of market risk sensitivities for the CVA portfolio and having supervisory approval of the methodology.

For CVA capital desks, the FRTB-CVA transition was one of the most operationally intensive regulatory changes of the Basel IV cycle. The calculation requires: computing regulatory CVA (distinct from accounting CVA) for each counterparty netting set; calculating CVA sensitivities to credit spreads, interest rates, FX, equity, and commodity risk factors; aggregating across risk factors using the Basel prescribed correlation structure; and recognizing eligible hedges to reduce the capital requirement. Claude helps with each of these layers — from explaining the framework to working through specific sensitivity calculations and capital comparisons. See also Claude AI for XVA Trading Desks and FRTB and Market Risk Capital with Claude for related workflows.

Regulatory CVA vs. Accounting CVA

A critical distinction: SA-CVA uses regulatory CVA, not accounting CVA (IFRS 13 / ASC 820 fair value). Regulatory CVA under FRTB is calculated using a specific formula prescribed by the Basel standard — it uses a different expected exposure methodology (SA-CCR-aligned, not simulation-based), a standardized discount factor, and excludes DVA. The accounting CVA, by contrast, may be calculated using Monte Carlo expected exposure profiles, may include DVA, and must reflect the bank's specific valuation methodology.

The formula for regulatory CVA under FRTB for a single netting set NS with a single counterparty c is:

CVA_NS = LGD_c × ∑ᵢ max(0, exp(−s_c·t_{i−1}/LGD_c) − exp(−s_c·t_i/LGD_c)) × (EEPE(t_{i−1}) + EEPE(t_i)) / 2 × DF_i

Where s_c is the counterparty credit spread (from CDS or credit risk mapping), LGD_c is 40% for senior unsecured counterparties by default (or 25% for covered bonds), EEPE(t) is effective expected positive exposure under SA-CCR methodology, and DF_i is the risk-free discount factor. This is simpler than the full simulation-based accounting CVA but must be computed consistently across all netting sets for the sensitivity calculation to work.

• "Regulatory CVA vs. accounting CVA reconciliation: We have a 5Y USD IRS netting set with a BBB corporate counterparty. Accounting CVA (IFRS 13 Monte Carlo): $485K. Regulatory CVA (FRTB formula): $340K. The gap is $145K. Walk me through the methodological sources of this difference: (1) how does FRTB's use of SA-CCR-based EEPE differ from our Monte Carlo expected exposure simulation for a standard interest rate swap, (2) why does the FRTB regulatory CVA formula exclude DVA while our accounting CVA is bilateral, (3) how does the discount factor treatment differ — FRTB uses risk-free, accounting uses OIS flat curve, (4) which is higher for in-the-money vs out-of-the-money netting sets and why, (5) are there any netting sets where regulatory CVA could exceed accounting CVA?"
• "FRTB regulatory CVA calculation for a simple netting set: Single counterparty, single netting agreement with: (1) pay-fixed 3Y EUR IRS, €30M notional, current MTM +€850K; (2) EUR/USD cross-currency swap, €20M/USD, 5Y remaining, current MTM −€320K. Counterparty CDS spread: 75bps (proxy from sector mapping). LGD: 40% (standard). SA-CCR EEPE: assume 3.2% of notional for the IRS and 5.8% for the XCCY given tenor and type. Calculate: (1) the FRTB regulatory CVA for each instrument, (2) the netting set aggregate CVA, (3) CVA sensitivities to a 1bp widening of the counterparty credit spread (CVA CS01), (4) CVA sensitivity to a 1bp parallel shift in the EUR risk-free curve (CVA IR01 delta)."

SA-CVA Sensitivity Calculation

The core of SA-CVA is computing delta and vega sensitivities of the aggregate CVA portfolio to market risk factors, then applying the Basel prescribed capital formula. The risk factors include: counterparty credit spreads (by tenor), risk-free interest rates by currency and tenor, FX spot rates (for cross-currency netting sets), equity prices, and commodity prices. Each CVA is sensitive to the risk factors that drive both the exposure profile (rates, FX) and the default probability (credit spreads).

The capital charge formula is: K_SA-CVA = ρ × K_delta + (1−ρ²)^(0.5) × K_vega where K_delta and K_vega are the delta and vega capital components, and ρ = 0.5 is the correlation between delta and vega risks. Each component aggregates across risk factor buckets using the prescribed within-bucket and cross-bucket correlation matrices. The Basel framework specifies risk weights (RW) for each credit quality bucket (e.g., IG vs. HY counterparties, by tenor) and correlations within and between buckets.

• "SA-CVA delta capital calculation for a credit spread risk factor: We have 12 counterparties in our CVA book. Aggregate CVA CS01 by credit quality and tenor bucket: IG <1Y: −$8K, IG 1-3Y: −$45K, IG 3-5Y: −$92K, IG 5-10Y: −$67K, HY <1Y: −$12K, HY 1-3Y: −$28K, HY 3-5Y: −$31K. (Negative means CVA increases when spreads widen — normal direction.) Using Basel SA-CVA risk weights: IG 3-5Y RW = 1.38%, HY 3-5Y RW = 2.75%. Apply the prescribed within-bucket aggregation formula with correlation ρ = 1.0 for same counterparty, 0.35 across IG counterparties, 0.35 across HY: (1) calculate the weighted sensitivity (WS_k = RW_k × s_k) for each tenor bucket, (2) aggregate within the IG bucket using the intra-bucket correlation formula: K_b = sqrt(sum_k WS_k² + sum_{k≠l} ρ_{kl} WS_k WS_l), (3) repeat for HY bucket, (4) aggregate across credit quality buckets using cross-bucket correlation γ = 0.35."
• "SA-CVA hedge recognition for a CDS hedging program: We have €8.5M of CVA CS01 in the 3-5Y IG bucket, concentrated in three automotive counterparties. Our hedge: long protection on iTraxx Main 5Y (index) — $100M notional, CS01 of the index at 3-5Y is approximately €6.2M per $100M notional. SA-CVA allows index CDS hedge recognition with a 0.5 'index decomposition' discount applied to the hedge sensitivity. Calculate: (1) the hedge-adjusted WS_k for the IG 3-5Y bucket after applying the index CDS hedge with the prescribed 0.5 discount, (2) the capital saving from the hedge recognition vs. running unhedged, (3) whether a single-name CDS on one of the three counterparties would be more capital-efficient per dollar of hedge cost, (4) what additional conditions must the hedge satisfy to qualify as an eligible CVA hedge under CRR3 Article 383c."

BA-CVA vs. SA-CVA: Capital Comparison

BA-CVA (Basic CVA) is the fallback for banks that can't or won't implement SA-CVA. It uses a simpler formula based on the regulatory CVA amounts directly — no sensitivity calculation, no hedge recognition (except for a "reduced BA-CVA" that recognizes single-name CDS hedges at a discount). The tradeoff: BA-CVA is operationally simpler but typically produces 2–4× higher capital requirements for banks with active CVA hedging programs. The capital comparison varies by portfolio type — for uncollateralized corporate loan books with no CDS hedging, BA-CVA and SA-CVA may produce similar outcomes; for dealer banks with large derivatives books and active CDS hedges, SA-CVA is materially cheaper.

• "BA-CVA vs. SA-CVA capital comparison for a representative portfolio: We have 85 counterparties in our CVA book, with aggregate regulatory CVA of $42M. Our CVA hedging program: $180M notional single-name CDS (roughly 60% of credit spread risk hedged), plus $500M iTraxx Main for systemic credit exposure. BA-CVA (full, no hedge recognition): uses the formula K_BA = α × sqrt((sum_c SCVA_c²) + (sum_{c≠d} ρ_cd SCVA_c SCVA_d)). Reduced BA-CVA (with single-name hedge recognition at 25% discount): eligible if we have a single-name CDS referencing each counterparty. SA-CVA (sensitivity-based, full hedge recognition): our CS01 net of hedges is approximately 35% of gross CS01. Estimate: (1) the relative capital difference between Full BA-CVA, Reduced BA-CVA, and SA-CVA for our profile, (2) which approach is most sensitive to the size of our index hedge, (3) at what gross CVA book size does the operational cost of SA-CVA implementation typically break even against capital savings?"
• "FRTB-CVA implementation decision for a regional bank: We are a regional UK bank with 45 corporate counterparties, mostly plain vanilla loan-associated hedges (IR swaps and FX forwards). Our derivatives book is €2.3B notional, primarily used to hedge corporate clients' balance sheet risk. We have no active CVA hedging desk — we currently take CVA as an accounting charge but do not hedge it externally. Under CRR3 (UK PRA rules effective 2025), we must calculate CVA capital under FRTB. Assess: (1) whether SA-CVA is viable for us given no active CVA hedging, (2) whether the reduced BA-CVA option (recognizing our limited single-name CDS hedges) is better, (3) what infrastructure we need to implement BA-CVA — data inputs, calculation engine, reporting framework, (4) what approximate capital impact (as % of current RWA) should we expect from the FRTB-CVA transition vs. the old CRR Article 384 CVA charge?"

Eligible Hedges and Recognition Rules

Determining which hedges qualify for CVA capital reduction under SA-CVA is an area where regulatory detail matters enormously — misclassifying an ineligible hedge as eligible causes restatement risk. The Basel IV rules (and CRR3 implementing rules for EU/UK) specify three categories of eligible CVA hedge: counterparty credit spread instruments (single-name CDS, single-name contingent CDS, equivalent), index CDS (recognized with a decomposition discount), and certain cross-currency instruments hedging FX-driven CVA exposure. Interest rate derivatives hedging the rates sensitivity of CVA are also eligible. Equity and commodity hedges are explicitly excluded.

• "Eligible hedge assessment for CVA capital purposes: We hedge our CVA book with the following instruments — provide a classification and eligibility ruling for each under SA-CVA: (1) 5Y single-name CDS on counterparty A (senior unsecured), bought protection; (2) 5Y CDS on counterparty A's parent company (not the legal entity in the netting set); (3) iTraxx Main 5Y index CDS, bought protection; (4) CDX NA IG 5Y, bought protection; (5) 3Y fixed-rate bond issued by counterparty A (held on our trading book, used as a proxy credit hedge); (6) 3Y EUR IRS received-fixed, used to hedge the interest rate sensitivity of CVA on our EUR-denominated netting sets; (7) USD/EUR cross-currency basis swap hedging the FX component of CVA on USD-denominated counterparties reported in EUR. For each eligible instrument, explain the documentation requirement under CRR3 and any discount applied to hedge recognition."

Data Requirements and Calculation Infrastructure

Implementing SA-CVA requires a calculation chain that starts with trade-level data and ends with aggregated risk factor sensitivities. The key data inputs are: full trade data for each netting set (notional, maturity, product type, currency); CSA terms (threshold, MTA, initial margin agreement); counterparty credit spreads by tenor (from single-name CDS if liquid, or sector/rating proxy mapping); risk-free yield curves by currency; SA-CCR inputs (replacement cost, PFE add-on by asset class); and CVA hedge portfolio data (reference entity, notional, tenor, CDS spreads). Building and maintaining this data chain — particularly the counterparty credit spread proxy mapping for non-CDS-traded corporate names — is typically the most operationally intensive part of FRTB-CVA compliance.

• "Credit spread proxy mapping for FRTB-CVA: We have 62 counterparties in our CVA book. Only 18 have actively traded single-name CDS. For the remaining 44, we need credit spread proxies. Our current approach: map each counterparty to the nearest iTraxx sector subindex using industry classification, and adjust for issuer-specific rating and maturity. FRTB-CVA allows proxy mapping but requires it to be 'reasonably reflective of the counterparty's creditworthiness.' Design a mapping framework: (1) what data do we need per counterparty to construct the proxy — rating, sector, geography, outstanding bond issuance, (2) how should we handle counterparties in sectors not covered by iTraxx subindices, (3) what is the prescribed credit spread proxy table in the Basel framework for counterparties with no CDS market — the 'unrated' bucket, (4) how should we validate that our proxies are within regulatory tolerance, (5) what documentation does the UK PRA require for credit spread proxy methodology under FRTB-CVA?"