Probability Interview Questions for Quant Roles
How to prepare for quant probability interview questions covering coins, dice, cards, Bayes, random walks, urns, and expected value.
Candidates who need a probability-first quant interview plan.
What probability interviews test
Probability rounds test whether you choose a model before you calculate. The surface objects map onto methods quite reliably: coins test independence and stopping rules, dice test counting and complements, cards test sampling without replacement, urns test conditioning, Bayes questions test base rates, and random walks test state recursion. Recognizing which of those six you are looking at, inside the first thirty seconds, does more for your score than any amount of formula recall.
Concrete example
Flip a fair coin until the first head appears - what is the expected number of flips? Write the recurrence: one flip always happens, then with probability 1/2 you are finished and with probability 1/2 you are exactly back where you started. So E = 1 + (1/2) x 0 + (1/2) x E, which gives E = 2. Summing n/2^n over all n produces the same 2, but the recurrence is faster and it generalizes: for a coin with head probability p the answer is 1/p, so a coin that lands heads 10 percent of the time takes 10 flips on average. Naming the recursive state before moving any symbols is the habit that transfers.
Work Bayes in counts, not percentages
The most reliable Bayes technique under interview pressure is to drop the formula and populate a table with counts. Take two urns: urn A holds 2 red and 3 blue balls, urn B holds 4 red and 1 blue. You pick an urn at random and draw a red ball - what is the chance it came from urn A? Imagine ten draws from each urn. Urn A produces 4 reds, urn B produces 8, so there are 12 reds in total and 4 of them came from A. The answer is 4/12 = 1/3. The formula returns the same number, but counts make the asymmetry visible, and it is far harder to misplace a denominator you can see written down.
How to practice
Build a method checklist and make yourself name the entry before computing: complement counting, conditioning on the first step, linearity of expectation, symmetry, Bayes in counts, and state recursion. Those six cover the overwhelming majority of interview probability. After each problem, write the single sentence that would have found the method faster - not "review cards" but "the hand is unordered, so the denominator is C(52,5)." Then run mixed sets, because recognition is the real skill and a set sorted by topic quietly removes the hardest part of the task.
Common mistakes
The dominant failure mode is treating probability as formula lookup rather than model selection. Close behind is assuming independence when draws happen without replacement - two aces off the top of a deck is (4/52) x (3/51) = 1/221, not (4/52)^2 - and mixing ordered with unordered counting inside a single calculation, which throws the answer out by a factorial. Slow down for the setup, then speed up the arithmetic once the structure is certain.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.