← BNP Paribas Interview Insights
This is the kind of question where you think you know it until someone asks you to actually put numbers on a portfolio.
Start by clearly defining the Greeks and their role in risk management, then focus on delta as the sensitivity of an option's price to changes in the underlying asset. For the portfolio delta, explain that it is the sum of each position's delta, and walk through a simple example with a few options and stock positions, computing the total delta exposure.
Pro tip: Emphasize that delta is not static—it changes with the underlying price (gamma) and time (charm), so a software engineer must consider real-time recalculation and efficient data structures for large portfolios.
Briefly explain that the Greeks measure sensitivity of option prices to various factors: delta (underlying price), gamma (delta), vega (volatility), theta (time), rho (interest rates).
Define delta as the rate of change of option price with respect to the underlying price, ranging from 0 to 1 for calls and -1 to 0 for puts. Mention that delta also approximates the probability of expiring in-the-money.
Explain that portfolio delta is the sum of deltas of individual positions, each weighted by the number of contracts and the contract multiplier (usually 100 for equity options).
Construct a simple portfolio: e.g., long 10 call options with delta 0.6, short 5 put options with delta -0.4, and long 100 shares of stock (delta 1 per share). Compute total delta: (10*0.6*100) + (-5*-0.4*100) + (100*1) = 600 + 200 + 100 = 900.
Interpret the total delta: a delta of 900 means the portfolio gains $900 for a $1 increase in the underlying. Discuss implications for hedging and the need for dynamic recalculation.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.
Knew the list: linearity, error independence, constant variance, normality of residuals, no perfect multicollinearity.
Start by listing the five standard assumptions of linear regression (linearity, independence, homoscedasticity, normality of residuals, and no multicollinearity). For each, briefly explain the assumption and then describe common detection methods, such as residual plots, statistical tests, and variance inflation factors. Emphasize that in practice, you would prioritize checking assumptions that most affect your use case and consider robust alternatives if violations are severe.
Pro tip: Mention that while formal tests exist, visual diagnostics (e.g., residual vs. fitted plots) are often more practical and interpretable, especially in large datasets where tests may flag trivial violations. Also, note that some violations (like non-normality) matter less for prediction than for inference.
Clearly state the five key assumptions: linearity, independence of errors, homoscedasticity, normality of errors, and no multicollinearity. Briefly define each in one sentence.
For linearity, suggest plotting residuals vs. fitted values or each predictor; for independence, use Durbin-Watson test or plot residuals vs. order of collection.
For homoscedasticity, use residual plots or Breusch-Pagan test; for normality, use Q-Q plots, histograms, or Shapiro-Wilk test on residuals.
Compute Variance Inflation Factor (VIF) for each predictor; VIF > 5 or 10 indicates problematic multicollinearity. Also, correlation matrix can help.
Briefly mention what to do if violations are found: transformations, robust standard errors, removing predictors, or using alternative models. This shows practical understanding.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.
These came interleaved with resume questions which threw me off a bit.
First, clarify the problem to ensure you understand the scenario and assumptions. Then, outline your solution step-by-step, explaining the reasoning and any formulas used. Finally, compute the answer and briefly discuss its implications or potential edge cases.
Pro tip: Verbalize your thought process clearly and check if the interviewer wants a quick estimate or an exact calculation. If time permits, mention how you would validate the result with a simulation or alternative method.
Ask questions to confirm the problem statement, assumptions, and what exactly is being asked. Ensure you understand terms like independence, distribution, and sample size.
State the probability or statistical concepts you'll use (e.g., conditional probability, Bayes' theorem, expected value) and why they apply. Sketch a step-by-step plan.
Carry out the math carefully, explaining each step. If numbers are complex, consider simplifying or using approximations, but note that you're doing so.
Interpret the result in the context of the problem and check if it makes sense (e.g., probability between 0 and 1). Mention any assumptions that could affect the answer.
If relevant, relate the problem to real-world scenarios or software engineering contexts, such as A/B testing, system reliability, or algorithm analysis.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.
Standard resume walkthrough but the follow-ups were sharp.
Choose a project that showcases both technical depth and business impact, ideally one involving trade-offs or root cause analysis. Structure your walkthrough using a clear narrative: context, problem, your specific actions, technical decisions, and measurable results. Be prepared to dive deep into any technical detail, explaining your reasoning and alternatives considered.
Pro tip: Proactively mention a trade-off you made and why, then briefly note what you would do differently now. This shows self-awareness and growth, which interviewers at banks like BNP Paribas value highly.
Briefly describe the project's purpose, your role, the team size, and the business domain. Keep it concise to leave time for technical depth.
Explain the specific technical challenge or goal, such as performance bottlenecks, scalability issues, or integration complexity. Highlight why it mattered to the business.
Walk through the steps you took, focusing on your individual contributions. Describe the technologies used and the rationale behind your choices.
Explain any key decisions where you weighed alternatives, and if applicable, how you diagnosed and resolved a critical issue. Show your analytical process.
Share measurable outcomes (e.g., reduced latency by 30%, increased throughput). Reflect on lessons learned and how you would improve the solution today.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.