How should a candidate prepare for a quantitative interview so the preparation resembles the work? This note suggests that preparation, using a plain question about the SPY spread as the public example, for internships and for experienced conversations in New York, London, and Hong Kong.
Practice explaining a limit
Quantitative interviews on this desk are conversations about reasoning under a stated assumption. They are not a hunt for a single numeric party trick. Preparation that only drills speed of arithmetic will feel busy and miss the sample. A better hour is spent taking a small public fact and saying, out loud, what it does not mean. The offices in New York, London, and Hong Kong are entitled to the same answer. A candidate who prepares a different personality for each city is preparing theater.
Internships use the same muscles with less assumed background. A summer candidate may be asked to reason about a coin, a queue, or a market convention and to notice when the question is underspecified. Guessing a distribution and charging ahead is the failure mode. Saying which fact is missing is the skill. That skill is also how a research note survives review. The interview is a short, artificial version of the note.
Preparation for a quantitative interview is practice at stating an assumption, a mechanism, and a limit. It is not memorization of a puzzle or of a market width.
The spread as a question, not a fact to recite
SPY is a liquid public fund. Its bid and its offer are usually close, and the distance between them is still a cost. A useful interview question is not the width on a particular afternoon. It is what a person thinks the width represents, who pays it, and when a last-trade price hides it. A candidate can prepare by reading the public definition of a spread and then explaining it without a prop. Memorizing a width would be a waste. The width changes, and the definition is the thing being tested.
Preparation then goes one step past the definition. What would make the candidate refuse to trust a backtest that ignored the spread? What would a point-in-time membership problem look like if the question moved from SPY itself to a basket? The candidate does not need the firm’s answers. The candidate needs the habit of attaching a limit to a computation. Interns are trusted, when they are trusted, to ship a tool that makes such a limit visible. They are not asked to arrive with a strategy. Experienced candidates are not asked to arrive with one either.
Geography does not change the question. A conversation in London may happen while New York is still in its session, and a conversation in Hong Kong may happen near a different open. The interviewer may mention that fact. It is context, not a separate syllabus. The SPY spread means the same thing in each office. Candidates who want to prepare by watching a market can watch one open and write down what they do not understand. That list is better material than a page of formulas unattached to a mechanism.
What not to prepare
Do not prepare a performance story, a guessed description of how the firm trades, or a figure for speed. The public notes do not support those answers, and an interview that rewards them is off the standard described here. Do not prepare a speech about why a particular city is the real office. New York, London, and Hong Kong are the three offices already on the site. Work is shared across them. An internship conversation and an experienced conversation differ in depth, not in the honesty they require.
How to prepare is to practice saying what a public mechanism does and does not imply. The SPY spread is a sufficient example. The rest is listening to the question that was actually asked, in whichever of the three offices the conversation happens to use.
This piece does not list interview questions, promise a process, or describe hiring numbers. It describes preparation that matches the work, for internships and experienced interviews in New York, London, and Hong Kong. The SPY spread is the example to practice explaining.
