PFE and Expected Exposure for Derivatives: Claude AI Prompts (2026)

Exposure Metrics in Counterparty Credit Risk

Counterparty credit risk management relies on a family of exposure metrics that serve different purposes: credit limit monitoring, CVA calculation, regulatory capital, and XVA pricing. Understanding the distinctions between PFE, EPE, ENE, and EEPE — and when each is the right metric — is foundational for derivatives risk desks. These calculations sit upstream of CVA (which uses EPE/EEPE as inputs), SA-CCR EAD (which uses its own standardized EEPE proxy), and SIMM IM (which is based on sensitivities, not exposure), making exposure methodology one of the most consequential choices in counterparty risk infrastructure.

Claude helps derivatives analysts and risk managers compute and interpret these metrics for specific portfolios. The examples below use simplified analytical approximations appropriate for plain vanilla products — for complex structured portfolios, full Monte Carlo simulation is standard practice, but the analytical approaches give good intuition and serve as sanity checks. For the broader counterparty credit risk context see XVA Trading Desk AI, and for SA-CVA capital uses of these metrics see SA-CVA under FRTB.

PFE Calculation for Interest Rate Swaps

For a plain vanilla IRS, the PFE profile over time can be approximated analytically. The exposure at time t is the MTM of the swap, which is the present value of the remaining fixed vs. floating cash flows. As rates move, this MTM changes — the exposure to a pay-fixed counterparty increases when rates fall (making the fixed payments more valuable to us). The PFE at time t at confidence level α is approximately:

PFE(t, α) = N × D(t, T) × σ_r(t) × Φ⁻¹(α) × √(t) × (T − t)

Where N is the notional, D(t,T) is the annuity discount factor from t to maturity T, σ_r(t) is the spot rate volatility (annualized basis point vol), Φ⁻¹(α) is the normal quantile at confidence level α, and (T−t) is the remaining swap life. This formula captures the key PFE features: PFE starts at zero (no uncertainty at initiation if ATM), rises to a peak around the midpoint of the swap life, then falls as the remaining term shortens.

• "PFE profile for a 5Y USD IRS: We have a $50M pay-fixed 5Y USD IRS, ATM at inception (fixed rate = current 5Y SOFR swap rate, 4.55%). Current spot: ATM, MTM ≈ 0. Rate volatility: 80bps annualized normal vol (approximately 18% relative vol at current rates). Calculate the 95th percentile PFE profile at t = 6M, 1Y, 2Y, 3Y, 4Y, 4.5Y: (1) apply the analytical approximation for each time step, (2) identify the peak PFE and at what time it occurs — is it at t = 0.5T, i.e., 2.5Y? (3) compare PFE to the current notional — what % of notional is peak PFE for this product type, (4) for credit limit purposes, should we use peak PFE or 95th percentile PFE at each horizon separately, (5) how does PFE change if the trade is 50bps in-the-money to us at initiation (current MTM +$1.1M)?"
• "PFE for an FX forward position: We have a 12M USD/EUR forward — we buy EUR 20M at the forward rate of 1.0820 (€1 = $1.0820). The FX forward PFE is driven purely by FX spot volatility (approximately 7.5% annualized). Unlike IRS, FX forwards don't have the hump-shaped PFE — for a single cash flow at maturity, exposure grows monotonically to maturity. (1) Calculate the 97.5th percentile PFE for this FX forward at 3M, 6M, 9M, and 12M, using the formula: PFE(t) = max(0, N × (F(t) − K) + N × σ_FX × Φ⁻¹(α) × √(T-t)) where F(t) is the current forward rate and K is the contract forward rate; (2) compare PFE to the current in-the-money amount if the EUR has moved to 1.10 (we are now $180K in the money); (3) explain why FX forward PFE at maturity equals approximately N × σ_FX × Φ⁻¹(α) × √T — the full lognormal quantile."

EPE and EEPE Calculation

EPE is the time-averaged expected positive exposure — the average over all future time steps of the expected positive MTM of the netting set at each step. Formally: EPE = (1/T) × ∫₀ᵀ E[max(V(t), 0)] dt, where V(t) is the netting set MTM at time t. For CVA calculations, EPE is used as the input exposure profile (quarterly or monthly), weighted by marginal default probabilities to get the CVA. EEPE (Effective EPE) is the regulatory version used in SA-CCR's IRB approach — it takes the running maximum of the EPE profile to prevent decreasing exposure profiles from gaming the calculation.

• "EPE calculation for a netting set with two offsetting swaps: We have two USD IRS trades with the same counterparty under a single ISDA netting agreement: (1) pay-fixed 5Y IRS, $50M notional, current MTM +$1.8M; (2) receive-fixed 7Y IRS, $30M notional, current MTM +$2.4M (same direction, both positive to us). Net MTM = +$4.2M. The netting set has two positive-exposure trades — there is no offset. Contrast with a different scenario: (3) pay-fixed 5Y IRS, $50M, MTM +$2.8M, and (4) receive-fixed 5Y IRS, $50M, MTM -$2.3M. Net MTM = +$0.5M. Calculate EPE for each scenario at 1Y, 3Y, 5Y horizons using the formula E[max(V, 0)] approximated as the Black formula for option-like exposure: EPE(t) ≈ V₀ × N(d₁) + σ_V × √t × φ(d₁). Show how netting reduces EPE in scenario 2 compared to scenario 1, and explain what determines the magnitude of the netting benefit."
• "EEPE construction for SA-CCR IRB approach: We have a netting set with quarterly EPE profile (annualized): Q1: $1.2M, Q2: $2.1M, Q3: $2.8M, Q4: $2.6M, Q5: $2.4M, Q6: $2.1M, Q7: $1.8M, Q8: $1.5M. (1) Construct the Effective EPE profile as the running maximum: EEPE(t) = max(EPE(s) for s ≤ t). Show the EEPE at each quarter. (2) Calculate the 1Y EEPE: the time-average of EEPE over the first year (Q1-Q4). This is the input to SA-CCR IRB for the first-year EAD. (3) Calculate the total EAD under SA-CCR IRB approach: EAD = α × EEPE₁Y where α = 1.4. (4) Compare this to the SA-CCR standardized EAD for the same portfolio — when would you expect them to differ materially?"

SA-CCR EAD Calculation

SA-CCR replaced the old Current Exposure Method (CEM) and is now the standard regulatory framework for counterparty credit risk EAD. The formula is: EAD = α × (RC + PFE_addon), where α = 1.4, RC is replacement cost, and PFE_addon is the potential future exposure add-on. The add-on is calculated by asset class (IR, FX, credit, equity, commodity), with each trade contributing to its asset class add-on through adjusted notional, supervisory delta, maturity factor, and supervisory factor. Trades can partially offset within a hedging set (same asset class and currency) but not across asset classes.

• "Full SA-CCR calculation for an IR derivatives netting set: Single netting agreement, no initial margin, threshold = $5M, MTA = $500K. Trades: (1) pay-fixed 5Y USD IRS, $50M notional, current MTM +$1.8M, 5Y residual maturity, ATM (delta ≈ 0.5); (2) receive-fixed 7Y USD IRS, $30M notional, current MTM +$1.2M, 7Y residual maturity, ATM (delta ≈ 0.5); (3) long USD 6M SOFR cap, $25M notional, 1Y to expiry, implied vol 90bps, delta 0.30. Step through the full SA-CCR calculation: (A) Replacement Cost: RC = max(V − C, 0) where V = net MTM = $3.0M, C = 0 (no collateral). RC = $3.0M. (B) Adjusted notionals: for IRS, d_i = notional × (MF) where MF = sqrt(min(M,1)/1). (C) Trade-level effective notional. (D) Supervisory delta for each trade. (E) Hedging set effective notional. (F) IR asset class add-on. (G) Total PFE add-on = multiplier × IR add-on. (H) Total EAD = 1.4 × (RC + PFE_addon)."
• "SA-CCR vs. CEM comparison for a large FX forward book: We have 40 USD/EUR FX forward contracts with a single corporate counterparty under one netting agreement. Total notional: $200M. Mix: approximately 60% short-dated (<1Y) and 40% medium-dated (1-5Y). Current net MTM: $3.5M in our favor. Under CEM (old method): EAD = V + 8% × notional for <1Y FX and 15% for 1-5Y FX, with a 40% netting benefit. Under SA-CCR: adjusted notional uses FX supervisory factor of 4%, maturity factor capped at 1 for <1Y trades, and the multiplier = min(1, 0.05 + 0.95 × exp(V/(2 × Aggregate_addon))). (1) Compute the approximate CEM EAD, (2) compute the SA-CCR EAD showing how the multiplier reduces it when current MTM is relatively small vs. the add-on, (3) for which portfolios would SA-CCR give higher EAD than CEM — and why does this happen for deep out-of-the-money netting sets?"

Collateral and Margin Period of Risk

The margin period of risk (MPoR) is the assumed time between the last successful margin call and the point at which the portfolio is fully hedged after a counterparty default. For standard bilateral OTC with daily margin calls, Basel III specifies a minimum MPoR of 10 business days for netting sets with fewer than 5,000 trades. The PFE and EPE calculation should be done over the MPoR, not the full trade life, when the trade is subject to daily margining — this dramatically reduces the effective exposure because losses beyond daily margin calls are bounded.

• "Margin-adjusted PFE for a collateralized netting set: We have a netting set with current MTM = $5.2M, and a CSA with: zero threshold (full two-way margining), MTA $500K, daily margin calls. Uncollateralized PFE over the trade life: $18M at 95th percentile at peak. Margin-adjusted PFE calculation over 10-day MPoR: (1) over 10 business days, how much can the MTM move for a portfolio with daily rate vol of 4bps and $300M DV01 aggregate? (2) calculate the 95th percentile MTM move over 10 days, (3) add the MTA and minimum transfer amount as additional exposure sources: effective PFE = MTA + MTA_counterparty + 10-day MTM shock, (4) compare the collateralized PFE ($X million) to the uncollateralized PFE ($18M) — what is the collateral benefit ratio, (5) how does the MPoR increase to 20 days for large netting sets (>5,000 trades) and why?"