Quant research interviews are not trick questions wearing a maths costume. They are a short, high-pressure sample of the work: can you take an unfamiliar problem, make it tractable, estimate an answer, and know how wrong you are. Most students over-prepare on content and under-prepare on process, which is the opposite of what the interview actually measures. The Optilic internship runs its first-round assessment on exactly this basis, and the pattern is consistent across the firms you have heard of.
This post is for maths, statistics, physics and quantitative CS students who have an interview coming up, or who are deciding whether to prepare for one. It assumes a semester or so of lead time. If you are preparing to apply to Optilic in particular, see the internship page for the current process; the advice below is general.
What they are actually testing
Three things, roughly in order of weight:
- Clean reasoning under uncertainty. Can you break a problem down, state assumptions, and reach an estimate that is defensible rather than confident?
- Calibration. Do you know which parts of your answer are solid and which are hand-wavy, and do you say so?
- Technical fluency. Probability, linear algebra, statistics, a little programming. These are the vocabulary, not the point.
We have written at length about the first two in Reading the assessment: what we actually mark. The short version: a wrong answer with a clean estimate and stated assumptions outranks a right answer that appears from nowhere.
What to study
Pick depth over breadth. The following list is small on purpose.
- Probability. Expectation, variance, conditional probability, Bayes. Common distributions and when each is a reasonable model. Random walks and basic Markov chains. The classic probability puzzle books (Mosteller, Blitzstein) are good; the goal is not to memorise them but to see enough variants that nothing feels alien.
- Statistics. Estimation, bias and variance, maximum likelihood, basic hypothesis testing, confidence intervals. Linear regression from first principles, including what the assumptions mean when they fail.
- Linear algebra. Projections, eigen-decompositions, PCA, and the geometric picture behind each. Being able to draw the thing matters more than being able to compute it by hand.
- One programming language, well. Python is the default. You should be able to load a CSV, clean it, fit a basic model, and plot the result without reaching for documentation on every line.
- Estimation. Fermi problems. Order-of-magnitude guesses. These show up in interviews because they are what the job is.
What to skip
Options pricing beyond the intuition. Stochastic calculus for its own sake. Memorised brainteaser answers. Finance vocabulary. None of these hurt, but none of them are the lever; spending a month on Ito calculus to look serious is a classic misallocation. If you need the tool for a question, the interviewer will usually walk you through the mechanics.
How to practise thinking out loud
The single biggest return on preparation time is practising spoken reasoning. Interviews reward process you can hear. Try this:
- Pick one problem a day. Set a timer. Record yourself solving it aloud.
- Play it back. Listen for the moments where you went silent, got stuck, or stated a guess as a fact.
- Rewrite the solution as a written answer. Ask: did my spoken version include the assumption I now rely on in writing?
Two weeks of this is worth more than two months of problem lists.
What the interview itself looks like
Most quant research interviews have three ingredients: a timed written or online assessment, one or two technical conversations, and sometimes a take-home project. The conversations are where candidates win and lose. Interviewers will push on your answer, not because it is wrong, but to see what you do when pushed. The correct behaviour is not to defend the answer; it is to re-examine it, out loud, and update if you were wrong. This is the single trait every quant research team is screening for, and it is almost never explicitly asked about.
On the day
Sleep. Water. Paper. State the problem back to the interviewer before you start, in your own words. If you are stuck, say what you have tried and what the obstacle is, rather than going silent. If you make a mistake and notice it, say so and fix it in the open; it is almost always the strongest moment of an answer.
Preparation does not need to be heroic. A semester of steady practice, honest reflection, and one or two mock interviews with someone who will tell you the truth will put you in a position where the interview is just a conversation about problems you already enjoy. If you would like that conversation to be with us, the Quantitative Finance Internship is open.