Most quant interview prep focuses on what to say — the right formula, the right code, the right framework. Almost none of it focuses on the subtle signals that quietly tell an interviewer "this candidate hasn't actually sat on a desk." Having sat on both sides of the table, the mistakes that sink strong candidates are rarely about knowledge. They're about instinct.
Here are seven red flags that come up again and again — each one small on its own, each one a tell.
1. Reciting Formulas Instead of Reasoning From Them
Ask a candidate to price a call option and they write Black-Scholes from memory, symbol by symbol. Ask why the formula has a discount factor on the strike but not the spot, and the room goes quiet. Interviewers aren't testing recall — calculators and cheat sheets exist. They're testing whether you understand a formula well enough to rebuild it if a term changes: what happens if the underlying pays a dividend, or rates go negative, or you need to price the same option under a different numeraire.
Fix: for every formula you memorize, be able to derive the one-line intuition behind each term. If you can't explain why $e^{-rT}$ is there, you've memorized, not understood.
2. No Sanity Checks on Your Own Numbers
A candidate computes an option's Delta as 1.4 and moves straight to the next step without blinking. On a real desk, an answer outside a sane range (Delta must be between -1 and 1 for a vanilla option, correlations between -1 and 1, probabilities between 0 and 1) is an immediate stop-and-check moment. Interviewers deliberately watch for this reflex — it's the single fastest way to tell if someone has actually worked with these numbers versus just solved textbook problems.
Fix: narrate your sanity checks out loud. "Delta should be bounded between 0 and 1 for a call, so let me double check this number" costs you five seconds and signals real desk instinct.
3. Treating Every Problem as Continuous and Frictionless
Textbook derivations assume continuous hedging, zero transaction costs, and infinite liquidity. Real markets have discrete rebalancing, bid-ask spreads, and market impact. A candidate who answers a hedging question with a pure Black-Scholes derivation and never mentions rebalancing frequency, slippage, or that discrete hedging leaves residual Gamma risk is describing a textbook, not a trading book.
Fix: after any continuous-model answer, add one sentence about what breaks in practice — transaction costs, discrete hedging error, or liquidity constraints. That one sentence usually separates candidates.
4. Silence Under Uncertainty Instead of Structured Guessing
Given an unfamiliar brainteaser or an estimation question with no clean answer, some candidates freeze rather than reason out loud. Interviewers are almost never grading the final number on an estimation question — they're grading whether you can build a defensible model live, state your assumptions, and adjust when challenged. Silence reads as an inability to operate without a known playbook, which is exactly the opposite of what a trading floor needs during a fast market.
Fix: narrate a structure even when you're unsure — "I don't know the exact answer, but here's how I'd bound it" is a stronger response than a long silence followed by a lucky guess.
5. Code That Works But Ignores Numerical Stability
A candidate implements Monte Carlo pricing that runs and returns a plausible number — but uses a naive finite-difference bump for Greeks with no thought to bump size, doesn't check convergence as paths increase, and never mentions variance reduction. It works on a happy-path test case. On a live coding round, the interviewer will usually push exactly here: "what if I bump too small?" or "how do you know this has converged?" Candidates who only tested the happy path have no answer ready.
Fix: before you say "done," mention convergence, bump-size sensitivity, or seed stability unprompted. It shows you've actually debugged numerical code before, not just written it once.
6. Confusing Correlation-Level Explanations With Causal Mechanisms
Asked why implied volatility skews downward for equity index options, a weak answer is "because puts trade more expensive than calls." That's a restatement of the observation, not an explanation. A stronger answer identifies the mechanism: leverage effect and crash-risk demand for downside protection. Interviewers use these "why" follow-ups specifically to separate candidates who've memorized market facts from those who understand the economic mechanism producing them.
Fix: for every market phenomenon you can describe, also know the one-sentence causal story behind it — not just that it happens, but why.
7. No Questions Back About the Desk's Actual Risk
At the end of an interview, a candidate who only asks about compensation, hours, or promotion timelines signals they're evaluating the job as a generic finance role. A candidate who asks "what's the biggest risk this desk actually carries right now?" or "how do you handle model risk when your calibration disagrees with the market?" signals they're already thinking like someone who will sit on that desk. This is one of the cheapest, highest-signal moves available and most candidates skip it entirely.
Fix: prepare two specific, desk-relevant questions in advance — about their book, their models, or their recent market challenges — not generic culture questions.
The Common Thread
Every one of these red flags comes down to the same gap: treating quant finance as a set of formulas to recall instead of a discipline of reasoning under uncertainty, noise, and real-world friction. The candidates who stand out aren't the ones with the most memorized material — they're the ones who reason out loud, sanity-check their own numbers, and never forget that every model is a simplification of a messier, adversarial market.