Optiver interview questions

How Optiver's process runs, as candidates and the firm describe it, and the questions candidates report, each with a worked solution. Free, no sign-up.

Practise these in the app Mock interview

How Optiver interviews

Sources: Glassdoor: Optiver OA components · QuantVault: Optiver process · Glassdoor: Optiver software engineer internship

Questions (17)

Try each one before opening the solution. Interviewers change the numbers, so learn the method.

Quantitative Trader · Technical · Markov chains · reported by a candidate

In a tennis game you're tied at 30–30. You win each point with probability 0.6.

  1. (a) What is your probability of winning the game?
    Answer and worked solution

    Answer: 9/13 ≈ 0.69231

    Key idea. 30–30 is strategically the same as deuce: you need two points in a row, and a split brings you back.

    1. Look at the next two points. Win both: p² = 0.36, you win. Lose both: q² = 0.16, you lose. Split: 0.48, back to 30–30 (or deuce).
    2. The splits just restart, so only the deciding pairs matter.
    3. P(win) = p²/(p² + q²) = 0.36/0.52 = 9/13 ≈ 0.692.

    Check. 0.692 > 0.6: needing two in a row amplifies the better player's edge.

    Trap. Trying to enumerate every path to the end of the game. The pairs trick makes it one line.

  2. (b) Now p = 0.9. What is your probability of winning?
    Answer and worked solution

    Answer: 0.9878

    Key idea. Same formula, p = 0.9.

    1. p² = 0.81, q² = 0.01.
    2. P = 0.81/0.82 ≈ 0.988.

    Check. The better player's edge is amplified further: a 90% point-winner almost never loses the game from 30–30.

Interviewers may push further: What is the expected number of points played from 30–30?

Source: Glassdoor (Nov 2018) · Try it in the app →

Quantitative Trading Intern · Interview · Combinatorics · reported by a candidate

You roll three dice at once and read them left to right.

  1. What is the probability the three numbers are strictly increasing?
    Answer and worked solution

    Answer: 5/54 ≈ 0.092593

    Key idea. Only rolls with three different values can be strictly increasing, and each such set has exactly one increasing order.

    1. Choose 3 different values: C(6, 3) = 20 ways.
    2. Each gives exactly one increasing roll.
    3. P = 20/216 = 5/54 ≈ 0.093.

    Check. Compare with 1/6 if the three numbers had to be different: P(all different) = 120/216, and 1/6 of those are increasing: 20/216. Same answer.

    Trap. Answering 1/6, which ignores repeated values.

Interviewers may push further: Non-decreasing instead? (C(8,3) = 56 multisets → 56/216.)

Source: Glassdoor · Try it in the app →

Trading intern · Interview · Probability · reported by a candidate

You roll a fair die six times.

  1. (a) What is the probability that every number 1–6 appears exactly once?
    Answer and worked solution

    Answer: 0.015432

    Key idea. Count favourable sequences over all sequences.

    1. All sequences: 6⁶ = 46,656.
    2. Each face exactly once: the orderings of 1–6, 6! = 720.
    3. P = 720/46,656 = 5/324 ≈ 0.0154.

    Check. Very unlikely: about 1 in 65.

  2. (b) Follow-up (ours): on average, how many rolls until you have seen every face?
    Answer and worked solution

    Answer: 14.7

    Key idea. Coupon collector: break the wait into stages, one per new face.

    1. With k faces seen, a roll is new with probability (6 − k)/6: wait 6/(6 − k).
    2. Sum over k = 0…5: 6(1/6 + 1/5 + 1/4 + 1/3 + 1/2 + 1) = 6 × 2.45.
    3. = 14.7.

    Check. The last face takes 6 rolls on average on its own, so the answer must be well above 6.

Interviewers may push further: How likely is it that you need more than 20 rolls?

Source: Glassdoor: Optiver intern interviews · Try it in the app →

Trader · Interview · Optimal stopping · reported by a candidate

Three face-down cards show n, n + 1 and n + 2, in random order, and you don't know n. Turn one over: keep it, or turn another. Keep that, or take the last card. You win the value of the card you keep.

  1. Under the best strategy, how much more than n do you win on average?
    Answer and worked solution

    Answer: 4/3 ≈ 1.3333

    Key idea. The first card tells you nothing, since n is unknown, but the second can be compared with it.

    1. Always look at a second card.
    2. If it's higher than the first, it's the middle or top card, equally likely: keep it (worth n + 1.5 on average, versus n + 1 for the last card).
    3. If it's lower, the last card is better on average: take it.
    4. Over the 6 equally likely orders this strategy wins 1, 2, 2, 2, 1 and 0 above n: 8/6, so 4/3 ≈ 1.33 above n on average.

    Check. Doing nothing clever gives n + 1 on average; the comparison adds a third.

Interviewers may push further: With four cards n … n + 3? Now you are told n is 0, 10 or 20, equally likely. Does the strategy change?

Source: Glassdoor: Optiver interview question · Try it in the app →

Trader · Online test · Mental maths · reported by a candidate

Examples from a report of Optiver's 80-in-8 arithmetic test (no calculator).

  1. (a) 46²
    Answer and worked solution

    Answer: 2116

    Key idea. Square a round number, then correct.

    1. 46 = 50 − 4.
    2. 2500 − 400 + 16 = 2116.

    Check. Ends in 6 (6² = 36).

  2. (b) 2/17 to four decimal places
    Answer and worked solution

    Answer: 0.1176

    Key idea. Know 1/17 = 0.0588235… and double it.

    1. 1/17 = 0.05882352941…
    2. × 2 = 0.1176470588…
    3. To four places: 0.1176.

    Check. 2/17 is a little under 2/16 = 0.125.

Interviewers may push further: Keep going: how many digits of 2/17 can you give in a minute?

Source: Glassdoor · Try it in the app →

Trader intern · Interview · Mental maths · reported by a candidate

Quick estimates, no calculator.

From a 2014 report; formats may have changed.

  1. (a) 44/157 in basis points, to the nearest 5
    Answer and worked solution

    Answer: 2802.5

    Key idea. Basis points: multiply the fraction by 10,000. Approximate the denominator, then correct.

    1. 44/160 = 0.275.
    2. 157 is about 1.9% less than 160, so 44/157 ≈ 0.275 × 1.019 ≈ 0.2803.
    3. ≈ 2,803 bp.

    Check. 44/157 is a bit more than 0.28, since 0.28 × 157 = 43.96.

  2. (b) 72.5 to the nearest integer
    Answer and worked solution

    Answer: 130

    Key idea. 72.5 = 7² × √7.

    1. √7 ≈ 2.6458 (2.65² = 7.02).
    2. 49 × 2.6458 ≈ 129.6.
    3. Nearest integer: 130.

    Check. Between 7² = 49 and 7³ = 343, much closer to 49 × 2.6.

Source: Glassdoor: Optiver trader internship · Try it in the app →

Quantitative Trader · Interview · Combinatorics · reported by a candidate

Six people of different ages sit at random around a round table.

  1. What is the probability that, starting from the youngest, ages increase going either clockwise or anticlockwise?
    Answer and worked solution

    Answer: 1/60 ≈ 0.016667

    Key idea. Rotations don't matter: fix the youngest person's seat.

    1. The other five can be arranged 5! = 120 ways.
    2. Exactly 2 are increasing (one clockwise, one anticlockwise).
    3. P = 2/120 = 1/60.

    Check. With 5 people it's 2/4! = 1/12, so this should be smaller.

Source: Glassdoor · Try it in the app →

Trader Trainee · Interview · Expected value · reported by a candidate

You flip five fair coins, winning $1 for each head and losing $1 for each tail.

  1. (a) What is your expected profit?
    Answer and worked solution

    Answer: 0

    Key idea. Linearity: the total is the sum of five coins.

    1. Each coin is +1 or −1 with equal probability: mean 0.
    2. Five coins: 0.
  2. (b) Follow-up (ours): what is the standard deviation of your profit?
    Answer and worked solution

    Answer: 2.2361

    Key idea. Independent variances add.

    1. One coin: (±1 − 0)² = 1, so variance 1.
    2. Five coins: variance 5, SD = √5 ≈ 2.24.

Interviewers may push further: What would you pay for the right to stop after any flip?

Source: Glassdoor · Try it in the app →

Graduate Trader · Interview · Statistics · reported by a candidate

You flip 100 fair coins.

  1. (a) What is the standard deviation of the number of heads?
    Answer and worked solution

    Answer: 5

    Key idea. Heads ~ Binomial(100, ½).

    1. Var = np(1 − p) = 100 × ¼ = 25.
    2. SD = 5.
  2. (b) Roughly what is the probability of exactly 50 heads?
    Answer and worked solution

    Answer: 0.0796

    Key idea. Approximate the binomial by a normal with σ = 5. The chance of one exact value is about the density there.

    1. Peak density of a normal: 1/(σ√(2π)) = 1/(5 × 2.507) ≈ 0.080.
    2. Exact: C(100, 50)/2100 ≈ 0.0796.

    Check. Even the most likely single outcome happens less than 1 time in 12.

    Trap. Answering ½ because 50 is "expected". Expected doesn't mean likely.

Source: Glassdoor · Try it in the app →

Junior Quant Trader · Interview · Expected value · reported by a candidate

You roll two dice and are paid their total, except that a double is cancelled.

The report says doubles are "cancelled" without saying how, so both readings are here.

  1. (a) If a cancelled double pays nothing, what is the game worth?
    Answer and worked solution

    Answer: 35/6 ≈ 5.8333

    Key idea. Start from the expected total of two dice and remove what the doubles contribute.

    1. E[total] = 7.
    2. Doubles (1,1) … (6,6) each have probability 1/36 and totals 2, 4, …, 12: contribution 42/36 = 7/6.
    3. E = 7 − 7/6 = 35/6 ≈ 5.83.
  2. (b) If you reroll after a double instead, what is it worth?
    Answer and worked solution

    Answer: 7

    Key idea. Rerolling doubles means conditioning on "not a double".

    1. E[total | not double] = (35/6)/(30/36) = 7.

    Check. Doubles average 7 too, so removing them doesn't change the mean.

Source: Glassdoor search: junior quant trader · Try it in the app →

Software Engineer Intern · Final round · System design · reported by a candidate

Design a low-latency trading system, from the exchange's market data arriving to your order reaching the exchange. Describe the components, what hardware they run on, and where the time goes. Go as deep as you can.

Model answer
  1. Market data in. Exchanges publish over UDP multicast, often on two redundant feeds (A and B). A feed handler decodes the exchange's binary protocol, checks sequence numbers, takes the first copy of each packet from either feed, and recovers gaps from a snapshot or retransmission service. Kernel-bypass networking (the NIC delivers packets straight to user space) avoids system calls and copies; some firms decode on FPGAs.
  2. Order book. Build the book per instrument from the updates: price levels in arrays indexed by tick for speed, best bid and offer kept current.
  3. Strategy. Compute fair values and decide quotes. The hot path is single-threaded and pinned to an isolated core: no locks, no memory allocation, no logging, data laid out for the cache. Other threads talk to it through lock-free single-producer single-consumer ring buffers.
  4. Risk checks before every order. Position and exposure limits, price bands (no order far from the market), order-rate throttles, and a kill switch. They must be in the path and fast.
  5. Order gateway. A TCP session per exchange using its binary order protocol or FIX: logon, heartbeats, sequence numbers, handling acknowledgements, fills and rejects, and resynchronising after a disconnect.
  6. Off the hot path. Asynchronous logging, monitoring and alerting, a drop copy of fills to reconcile positions, and recording of all data for replay and testing.
  7. Hardware and placement. Servers colocated in the exchange's data centre, fast NICs with hardware timestamps, clocks synchronised by PTP, and tuned machines (isolated cores, no CPU frequency scaling, huge pages).
  8. Where the time goes. Wire to decoded update, to decision, to risk check, to order on the wire: measure each stage with hardware timestamps and watch the tail (the 99.9th percentile), not just the average. A cache miss is about 100 ns and a system call around a microsecond, so they show up.

Say what you would build first, and what happens on failure: a gap on the feed, a disconnect from the exchange, a strategy behaving badly (the kill switch). Ask what latency matters for this business before optimising.

Source: Glassdoor: Optiver software engineer internship (2025): "designing a high-frequency trade system … software and hardware components in detail" · Try it in the app →

Software Engineer Intern · Final round · System design · reported by a candidate

Design a system for selling concert tickets. A popular show sells out in minutes, and no seat may ever be sold twice.

Model answer
  1. Requirements first. No double-selling (correctness over speed), fair access when demand spikes, people pick seats, payment can fail. Ask about scale: seats per show, buyers at the start of a sale.
  2. Data. Events, seats (event, section, row, number, status: free, held, sold), holds (seat, buyer, expiry) and orders.
  3. Hold, then pay. Selecting a seat places a hold for a few minutes. The hold must be atomic: one conditional update such as "set status to held where id = ? and status = free", or a transaction with a row lock. Exactly one buyer's update succeeds. Holds that expire release the seat.
  4. Payment. Charge with an idempotency key so a retry never charges twice; on success mark the seats sold and the hold finished, in one transaction.
  5. The spike. A virtual waiting room lets buyers in at the rate the system can take. Seat maps are read far more than written, so cache them and accept being a second stale; the hold itself always goes to the database.
  6. Scaling. Shard by event, since one event's seats are only contended among its own buyers. Keep services stateless behind a load balancer.

Then talk about failure: what happens if a server dies after the payment succeeds but before the seats are marked sold? (Reconcile from the payment provider's records and the order log.)

Source: Glassdoor: Optiver software engineer internship (2024): "Design a ticketing system for buying concert tickets" · Try it in the app →

Software Engineer Intern · Technical · Data structures · reported by a candidate

Talk me through how you would implement a queue. Now build it on an array. What does each operation cost? What changes if it has to grow, or if one thread pushes while another pops?

Model answer
  • Linked list: keep head and tail pointers; push at the tail, pop at the head, both O(1). Each node is a separate allocation, which is slow and scatters memory.
  • Array (ring buffer): keep the array, the index of the front and the count. Push writes at (head + count) mod capacity; pop reads at head and moves it on. Both O(1), nothing shifts. Store the count (or leave one slot empty), or you can't tell full from empty.
  • Growing: when full, allocate double the size and copy the items in order to the start. Pushes stay O(1) amortised, because the copies add up to less than 2n.
  • Two threads: the simple answer is a mutex (plus a condition variable to wait on). With exactly one producer and one consumer you can avoid locks: only the producer writes the tail index and only the consumer writes the head, both atomics with release and acquire ordering, and each index on its own cache line so the cores don't fight over it (false sharing). Trading systems pass messages between threads this way.

Then write it: practise in the Coding section with a double-ended queue on a fixed array.

Source: Glassdoor: Optiver software engineer intern (2024–2025): "Talk through how to implement a queue … follow-up is to actually build an array-based queue" · Try it in the app →

Software Engineer Intern · Technical · CS fundamentals · reported by a candidate

What is the difference between TCP and UDP? Which would you use to receive market data, and which to send orders, and why?

Model answer
  • TCP: a connection (three-way handshake: SYN, SYN-ACK, ACK) carrying a reliable, ordered byte stream. Lost segments are retransmitted, and flow and congestion control slow the sender down. A lost segment holds up everything behind it (head-of-line blocking). Nagle's algorithm can delay small writes; low-latency code sets TCP_NODELAY.
  • UDP: independent datagrams with no connection, no retransmission and no ordering. Less overhead and no waiting for lost data, and it supports multicast.
  • Market data: UDP multicast. The exchange sends each update once to every subscriber. Stale prices are useless, so don't wait for a retransmission: detect the gap from sequence numbers, mark the book as unreliable, and recover from a snapshot or a separate retransmission channel (often TCP).
  • Orders: TCP. Every order, cancel and fill must arrive exactly once and in order, and the exchange needs a session for each firm. Exchange order protocols run over TCP.

Source: Glassdoor: Optiver software engineer intern (2024, 2026): "Difference between UDP and TCP" · Try it in the app →

Software Engineer Intern · Technical · CS fundamentals · reported by a candidate

Compare Python, C++ and Java. Why might a trading firm write its fastest systems in C++ and its tools in Python?

Model answer
  • How they run: C++ compiles ahead of time to machine code. Java compiles to bytecode that a JIT compiles while the program runs, so it needs warming up. CPython interprets bytecode, roughly 10 to 100 times slower for plain loops.
  • Memory: C++ is managed by hand, made safe with RAII (destructors free resources at the end of a scope) and smart pointers, and you control the layout of data in memory. Java and Python use garbage collection (Python mainly reference counting), so there are pauses or overhead you don't control directly.
  • Types: C++ and Java are statically typed; Python is dynamic, which makes it fast to write and easier to get wrong at run time.
  • Concurrency: C++ and Java threads run in parallel; CPython's global interpreter lock stops threads running Python code at the same time (processes or native libraries such as NumPy get round it).
  • The trade-off: on the latency-critical path, predictability matters most: no garbage collection pauses, no JIT warm-up, control of the cache. That favours C++. For research, analysis, tooling and glue code, speed of writing matters more, and Python's libraries win. Java sits between, with careful tuning to avoid garbage collection on the hot path.

Source: Glassdoor: Optiver software engineer interviews (2024–2025): "Difference between Python and C++", "Java vs C++" · Try it in the app →

Software Engineer Intern · Interview · Behavioural · reported by a candidate

Behavioural questions candidates report from Optiver's software engineering interviews:

Answer one out loud, as you would in the interview.

Model answer

Why Optiver rather than a tech company: the software is the business. Its performance and correctness show up straight away in profit and loss, engineers sit with traders, and problems such as low latency and risk are unusual. Name something concrete you learned from Optiver's own material or from someone there.

What a market maker does: it quotes both a price to buy (bid) and a price to sell (ask) in many products, all the time, so others can trade when they want. It earns the spread between them and manages the risk of the positions it ends up with, hedging quickly. The skill is pricing accurately and reacting fast, which is why the technology matters.

Projects: be ready to go deep: why you chose that design, what you would change, what broke and how you found it. Only claim what you can defend line by line.

Failure and weakness: Use a short story: the situation in one sentence, what you did (not the team), the result with a number if you have one, and what you'd do differently. Aim for about two minutes. Pick a real weakness with a cost, and what you now do about it.

Source: Glassdoor: Optiver software engineer intern interviews (2023–2026) · Try it in the app →

Trader · Interview · Behavioural · reported by a candidate

Behavioural questions candidates report:

Answer one out loud, as you would in the interview.

Model answer

Three parts, about a minute each. Why trading: a specific moment you got hooked (a game, a market you followed, a decision under uncertainty), not "I like maths". Why this firm: something concrete about how it trades or trains people, ideally from talking to someone there or from its own material. Why you: evidence you already think in probabilities and react well to being wrong. Avoid anything about money or prestige.

For the others: Use a short story: the situation in one sentence, what you did (not the team), the result with a number if you have one, and what you'd do differently. Aim for about two minutes.

Source: Glassdoor: Optiver interviews (QT intern, Oct 2026) · Try it in the app →

Other firms

Jane Street · SIG (Susquehanna) · IMC · Citadel / Citadel Securities · Five Rings · Two Sigma · Flow Traders · Maven Securities · Wincent · Akuna Capital · Da Vinci Derivatives · DRW · Common formats · Hudson River Trading · Jump Trading · Eclipse Trading · Old Mission · Belvedere Trading · Squarepoint Capital · Valkyrie Trading · Virtu Financial · Tibra · Tower Research Capital · D. E. Shaw · Mako · Question library

Practise by subject instead: topic guides.