XVA interviews test whether you can connect pricing, exposure, counterparty credit, funding, collateral, and simulation. Memorizing an integral for CVA is not enough. You need to explain what is simulated, where netting enters, how default and exposure interact, and which assumptions drive the number.
Use a layered answer: trade and agreement mechanics, future exposure, probability and discounting, adjustment definition, numerical implementation, sensitivities, and controls. That structure remains useful as the interviewer changes products or asks about production.
What this guide helps you do
- Explain the exposure engine that sits beneath XVA metrics.
- Distinguish CVA, DVA, FVA, MVA, KVA, and collateral effects.
- Describe netting, collateral, closeout, and wrong-way risk clearly.
- Discuss numerical architecture, hedging, and model controls.
1. Build the future exposure picture first
Future exposure is a distribution of portfolio values across future dates and market scenarios, after applying the relevant netting and collateral terms. Positive exposure represents potential loss if the counterparty defaults and does not pay; negative exposure is relevant to the institution's own default and funding view.
Expected exposure is a mean profile, potential future exposure is a quantile profile, and expected positive exposure summarizes positive values over time under a defined weighting. Always state measure, netting set, collateral assumptions, time grid, and treatment of exercise or path dependence.
- Simulate correlated market factors on a future time grid.
- Revalue or approximate every trade in the netting set at each state.
- Apply collateral calls, thresholds, minimum transfer amounts, lag, and margin period of risk.
- Aggregate positive and negative exposure consistently with closeout terms.
2. Distinguish the major valuation adjustments
CVA reflects the expected discounted loss from counterparty default, typically combining positive exposure, loss given default, and default probability. DVA is the symmetric own-default component under the selected valuation framework. FVA relates to funding costs and benefits, while MVA focuses on the cost of funding initial margin. KVA represents the cost of capital, and ColVA captures collateral remuneration effects in some decompositions.
Definitions and allocation conventions differ across institutions. Say which framework you are using, avoid double counting, and explain interactions rather than presenting every adjustment as an independent add-on.
| Adjustment | Primary driver | Key question |
|---|---|---|
| CVA | Positive exposure, counterparty default, recovery | How are exposure and default dependence modeled? |
| DVA | Negative exposure and own default | How is own credit represented and interpreted? |
| FVA | Funding requirement and funding spread | What is funded, and how is double counting avoided? |
| MVA | Initial-margin profile and funding cost | How is future initial margin estimated efficiently? |
| KVA | Capital profile and target return | Which capital measure and allocation are used? |
3. Treat legal agreement terms as model inputs
Netting can reduce exposure because positive and negative trade values offset within an enforceable netting set. Collateral adds path-dependent operational rules: thresholds, independent amounts, call frequency, eligible assets, haircuts, disputes, settlement lag, and closeout period.
A trade-level CVA calculation that ignores the agreement can materially misstate risk. Explain how trades map to counterparties and agreements, how legal enforceability is represented, and what happens when agreement data is missing or inconsistent.
4. Explain wrong-way risk and dependencies
Wrong-way risk occurs when exposure tends to increase as counterparty credit quality deteriorates. General wrong-way risk comes from broad economic dependence; specific wrong-way risk comes from a direct structural link, such as exposure to an asset closely tied to the counterparty.
A deterministic exposure profile multiplied by marginal default probabilities misses this dependence. Possible treatments include stressed parameters, copulas, joint factor-credit models, scenario overlays, or conservative add-ons. The choice depends on data, materiality, and use.
5. Discuss implementation, sensitivities, and controls
Full revaluation across many paths, dates, and trades is expensive. Production engines use regression, proxies, grids, adjoints, path reuse, netting-level aggregation, and distributed computation. Every approximation needs convergence, benchmark, stability, and reconciliation evidence.
XVA sensitivities support hedging and explain PnL, but may be noisy because market dynamics, credit curves, collateral, and simulation interact. Common controls include trade population reconciliation, curve and agreement-data checks, independent benchmarks, path and time-grid convergence, sensitivity bump tests, and unexplained-PnL thresholds.
- Separate market simulation, trade valuation, aggregation, and adjustment integration for testing.
- Version trades, agreements, curves, model configuration, and random seeds.
- Benchmark simplified portfolios with deterministic or analytical expectations.
- Monitor coverage, runtime fallbacks, approximation error, and reconciliation breaks.
Practise aloud
Interview drills with answer direction
Question 1
What is CVA?
Answer direction: CVA is the adjustment for expected discounted counterparty default loss, driven by future positive exposure, default likelihood, recovery, discounting, and their dependence under the chosen framework.
Question 2
What is the difference between CVA and PFE?
Answer direction: CVA is a valuation adjustment integrating expected default loss over time. PFE is a high quantile of the future exposure distribution at a date or horizon and is a risk measure, not a price adjustment.
Question 3
How does netting affect CVA?
Answer direction: Values offset within an enforceable netting set before positive exposure is taken. Because the positive-part operation is nonlinear, netting-set CVA is not generally the sum of standalone trade CVAs.
Question 4
What is wrong-way risk?
Answer direction: It is adverse dependence between exposure and counterparty credit quality, so exposure is high when default is more likely. Explain whether the dependence is general or counterparty-specific and how it is modeled or controlled.
Question 5
How would you validate an XVA engine?
Answer direction: Review theory and agreement mapping; test market simulation, valuation, collateral, default curves, aggregation, and integration separately; benchmark simple portfolios; run path and grid convergence; reconcile populations and sensitivities; assess limitations.
Turn reading into practice
A focused study plan
- Layer 1
Trades and agreements
Review cash flows, optionality, netting sets, CSAs, collateral mechanics, and closeout.
- Layer 2
Exposure
Build future value distributions and distinguish EE, EPE, PFE, and negative exposure.
- Layer 3
Adjustments
Derive the drivers and interactions of CVA, DVA, FVA, MVA, KVA, and ColVA.
- Layer 4
Production
Practise simulation architecture, approximations, sensitivities, validation, and controls.
Self-review
Frequent mistakes to catch early
- Defining CVA without the netting, collateral, recovery, and dependence assumptions.
- Confusing expected exposure with a PFE quantile.
- Treating agreement terms as operational detail rather than exposure inputs.
- Adding XVA components without checking interactions and double counting.
Continue with structured practice
Relevant Desk2Quant resources
XVA Calculus Lab: Counterparty Credit Risk
Work through CVA, DVA, FVA, MVA, KVA, ColVA, sensitivities, and counterparty-risk calculations.
Explore this resourceTrade Lifecycle for Quants
Understand where XVA enters booking, valuation, PnL, collateral, risk, and validation workflows.
Explore this resourceCredit Models: Quant Interview Playbook
Deepen default modeling, recovery, credit curves, migration, and wrong-way-risk intuition.
Explore this resourceGreeks, Vols, Yield Curves, Numerical Methods, Monte Carlo & XVA Guide
Connect XVA sensitivities to volatility, curves, Monte Carlo, and daily desk diagnostics.
Explore this resourceKeep building
Related quant finance guides
Common questions
Frequently asked questions
What topics appear in an XVA interview?
Expect future exposure simulation, CVA, DVA, FVA, MVA, KVA, netting, collateral, default curves, recovery, wrong-way risk, sensitivities, numerical methods, controls, and trade lifecycle questions.
What is the simplest way to explain CVA?
CVA is the present value of expected loss from counterparty default. It combines future positive exposure, probability of default, loss given default, discounting, and any dependence between exposure and credit quality.
Why is XVA computationally expensive?
Many trades must be valued across many future dates and correlated scenarios, then aggregated under agreement rules and integrated with credit, funding, margin, or capital quantities. Sensitivities multiply that workload.
Do XVA interviews require coding?
Often yes. Python, C++, SQL, numerical methods, simulation, data handling, and performance can all appear, especially for implementation, analytics, or model validation roles.