Risk analyst interview questions and how to answer them
A risk analyst interview tests two things at once: whether you can build the number, and whether you'll defend it when a desk head says it's wrong. Expect a technical grilling on the measures your team reports, a data exercise, and at least one question about pushing back on someone senior. The panel is picturing you on a bad morning, so give them answers that sound like you've had one.
What happens in each round
- 1
Recruiter screen
What happens
Which kind of risk you're aiming at. Market, credit, operational and model risk teams hire under the same title, so say which one you want and why. A fuzzy answer here gets you sent to the wrong hiring manager.
- 2
Technical interview with a senior analyst
What happens
Whether you understand the measures, not just the names. They'll ask you to explain VaR, expected loss or a control test in plain words, then push on the assumptions until they find the edge of what you know.
- 3
Data or Excel exercise
What happens
You get a messy file of positions or loans and a short window to summarise exposure, spot the bad records and write a few lines on what you found. They care more about the bad records you caught than how pretty the pivot table looks.
- 4
Hiring manager and a partner from the business
What happens
Whether you can challenge a trader, a relationship manager or a process owner without starting a war. They're also checking your writing, since your summary will end up in front of a risk committee.
Questions you're likely to get
1.Explain value at risk to me as if I sat on the board and had never heard of it.
Why they ask
Market risk teams report VaR every day, and half the job is explaining it to people who don't model anything. They want to hear whether you can make it plain without making it wrong.
How to answer
- Say it's an estimate of how much a portfolio could lose on a normal bad day, at a chosen confidence level and time horizon
- Name the three common methods: historical simulation, parametric and Monte Carlo, in one line each
- Be clear about what it doesn't tell you: how bad the loss gets beyond that threshold
- Mention expected shortfall and stress tests as the tools that cover that gap
2.A desk is over its limit this morning and the desk head says the model is wrong. What do you do?
Why they ask
This happens on real risk teams more weeks than it doesn't. They want to see that you check before you escalate, and escalate anyway when you should.
How to answer
- Rebuild the exposure from the raw position data to confirm the breach is real and not a feed problem
- Ask the desk for specifics on what they think the model misses, then test that claim
- Log the breach and escalate under the limit policy on time, even while the argument continues
- Agree a plan and a date to get back inside the limit, and follow up on that date
3.What's the difference between probability of default, loss given default and exposure at default?
Why they ask
Credit risk roles build expected loss from these three pieces, and CECL and IFRS reserve work sits on top of them. It's a quick filter for whether you've touched credit modeling at all.
How to answer
- PD is the chance a borrower defaults over a set horizon
- LGD is the share of the exposure you lose after collateral and recoveries
- EAD is how much is owed at the moment of default, including undrawn lines that get drawn
- Expected loss is the product of the three, and each has its own data problems
4.You pull loan data for a regulator request and a chunk of records have a blank property type. Walk me through what you'd do.
Why they ask
Bad data is the most common thing a risk analyst deals with, and a regulator deadline makes it worse. They want to see judgement, not just a SQL fix.
How to answer
- Size the gap first: how many loans and how much exposure are affected
- Try to fill the field from another source, such as the loan file, the appraisal or the collateral system
- Tell your manager early if the deadline is at risk, and don't guess values to make the table look complete
- Disclose any remaining gap in the submission and raise a data quality issue so it gets fixed at the source
5.How would you design a stress scenario for a commercial real estate loan book?
Why they ask
Stress testing is a big part of credit and enterprise risk work, and they want to know if you can link a story about the economy to losses on specific loans.
How to answer
- Start with a plausible story: rising rates, falling office occupancy, tighter refinancing
- Translate it into drivers for each loan, like property values, rent income and debt service coverage
- Flag loans that breach covenants or can't refinance at maturity under the scenario
- Explain how you'd sanity-check the result against past downturns and the team's previous runs
6.How do you test whether a control actually works?
Why they ask
Operational risk analysts spend a lot of time on control testing. This shows whether you understand the gap between a control on paper and one that happens.
How to answer
- Separate design, whether the control would catch the risk if done properly, from operating effectiveness
- Pick a sample of real instances and check for evidence, such as a sign-off, a reconciliation or a system log
- Talk to the person who performs it, because the procedure document is often out of date
- Rate any failure, agree an action with the owner and track it to closure
7.Tell me about a time your analysis showed something people didn't want to hear.
Why they ask
Risk is an independent function, and you'll regularly bring bad news to people who outrank you.
How to answer
- Pick a real case with a named audience, such as a manager, a client team or a professor
- Show how you checked your work before you raised it
- Describe how you framed it: facts first, then options, not blame
- Say what happened next, including if they didn't act and what you learned
8.What does backtesting a model tell you, and what does it miss?
Why they ask
Market and model risk teams backtest regularly, and regulators ask about it. They want to know you can read an exception count sensibly.
How to answer
- Backtesting compares predicted losses or limits with what actually happened
- Too many exceptions suggest the model underestimates risk; too few can mean it's overly cautious
- It won't catch risks that simply haven't shown up in the test window
- Pair it with benchmarking and a review of assumptions rather than trusting it alone
9.Write me a SQL query that shows total exposure by counterparty, only for counterparties above a threshold.
Why they ask
Almost every risk team pulls from a warehouse, and some interviewers will make you write it on a whiteboard or in a shared editor.
How to answer
- Use GROUP BY on the counterparty with a SUM of exposure
- Filter the groups with HAVING, not WHERE, and explain why
- Mention joins to a counterparty table for names and parent entities
- Point out that netting and parent-level aggregation can change the answer
10.What risk do you think is under-watched in our industry right now?
Why they ask
They want to see if you read beyond your coursework and can form a view.
How to answer
- Pick one risk you can actually explain, such as concentration in office property, third-party vendor outages or model drift
- Say how it would show up on this team's reports
- Suggest one indicator you'd track to catch it early
- Keep it short and invite them to push back
11.How would you write a one-page summary of a breach for the risk committee?
Why they ask
Committees read the note and skip the chart. Your writing decides whether anyone acts.
How to answer
- Lead with what happened and whether it's resolved
- Give the size of the breach against the limit and the cause in plain words
- State the action, the owner and the date
- Leave method detail for an appendix
Mistakes that sink good candidates
Describing risk as the team that blocks deals
Say you help the business take risk it understands.
Reciting textbook definitions with no example of using them on real data
Saying you'd fix a data gap by filling in plausible values, even as a shortcut
Being vague about which risk type you want, which reads as not knowing what the job is
Need more Risk Analyst interviews to prep for?
HeroApply applies to Risk Analyst jobs that match you, every day. 433 jobs are open today.