HackTheRounds Interview Experiences

SIG Quant Trader Interview Experience (2026) - 17-Question OA with Probability, Dice & Balance Puzzles, Pending

Susquehanna SIG Quant OA: 17 questions in 60 minutes, covering legs in barn linear systems, round table logic, Bayesian factories, three dice expected value, ba

By Anonymous · 2026-04-07

Background

Susquehanna's Quant OA is the fastest hour of my interview cycle so far. Seventeen questions, sixty minutes, no partial credit visible to the candidate, and the question style sits exactly where probability meets mental arithmetic meets logic puzzles. I took it last week as a QT candidate out of a stats masters program, and the difficulty was about what I expected from reading prior-year reports: the pool is stable but each cycle they rotate in small framing changes to stop rote memorization. Result is still pending. Writing this now because the structure is worth knowing cold before you sit down.

Timeline

OA Format (60 min, 17 questions)

SIG's Quant OA is delivered on a custom proctored platform. You see all seventeen questions up front and can skip between them, but a strict countdown means you cannot budget more than three and a half minutes per question on average. The question mix across my sitting was roughly: six probability, five logic puzzles, four quick-arithmetic or balance-system problems, and two word-problem setups that reduce to systems of equations. The fastest candidates in my prep group finished with five minutes to spare. I finished with under one.

Representative Question 1: Legs in the Barn

Problem: You enter a barn and see spiders, chickens, and cows. There are 520 legs in total. The number of chickens is twice the number of cows. The number of spiders is twice the number of chickens. Find the number of each.

Straight linear algebra. Let c be cows (4 legs), chickens is 2c (2 legs), spiders is 4c (8 legs). Total legs: 4c + 2 2c + 8 4c = 4c + 4c + 32c = 40c = 520 , so c = 13 , chickens 26, spiders 52. Answer. The time trap is doing the arithmetic in your head without writing down the coefficient sum, which is what turns a 45-second problem into a two-minute problem.

This question is coming soon to HackTheRounds.

Representative Question 2: Round Table Seating

Problem: Five toddlers (Anna, Brian, Charlie, Dixie, Eva) sit at a round table with constraints. (i) Anna will not sit next to Brian or Eva. (ii) Brian will not sit next to Charlie. (iii) Dixie will not sit next to Eva or Charlie. If Dixie is to the left of Anna, who is to the left of Brian?

Logic puzzle. Fix the rotation by placing Dixie and then Anna clockwise. With Anna not next to Brian or Eva, Brian and Eva cannot be the neighbors at positions two seats away from Anna on either side. Walk through the remaining five positions and propagate the adjacency constraints. The technique is constraint propagation, not enumeration. I solved it on paper under a minute by sketching the circle and crossing out invalid adjacencies. On the SIG UI there is no scratch paper built in, so I used the browser's developer-console memo pad. Fastest candidates do this on paper next to their keyboard.

Representative Question 3: Conditional Probability with Age Group Table

Problem: A table shows exercise preferences by age group across 1000 people. If you meet a 33-year-old who took the survey, compute the probability they prefer swimming.

Conditional probability via the table row for the relevant age bucket. Divide the swimming count in that bucket by the total count in that bucket. The trap is misreading "33-year-old" as the entire 30-39 band rather than the specific row the problem's table uses (sometimes 30-40 inclusive, sometimes 25-35, depending on the problem's table setup). Check the table's bin boundaries carefully.

This question is coming soon to HackTheRounds.

Representative Question 4: Balanced System Weight

Problem: A balanced system labeled 96 pounds in total contains several labeled shapes arranged across balance beams. Compute the weight of the green triangle.

Linear systems, solve with substitution. On my sitting the shapes were triangle, circle, square, each repeated some number of times on each side of two balance beams. Write one equation per beam (left weight equals right weight), add the constraint that total weight is 96, and solve. Three equations, three unknowns. The trap is missing that the outer balance beam weighs the inner balance beams as blocks, which propagates the shape counts multiplicatively.

Time budget: 2.5 minutes. If you cannot set up the system in under a minute, move on and come back.

This question is coming soon to HackTheRounds.

Representative Question 5: Widget Factory Bayes

Problem: Factory A produces 40% red widgets, 60% black. Factory B produces 80% red, 20% black. You select a factory uniformly at random, then sample two widgets uniformly at random from that factory. Compute the probability the factory was A given the two widgets drawn.

Bayesian update. The trick is that the question does not specify the colors of the two sampled widgets. Re-read: SIG's actual prompt often gives the two observed colors, and the answer depends on which specific pair. For two red widgets, the likelihood ratio is (0.4^2 0.5) / (0.4^2 0.5 + 0.8^2 0.5) = 0.16 / (0.16 + 0.64) = 0.2 , so P(A | two reds) = 0.2 . For two black, P(A | two blacks) = (0.6^2) / (0.6^2 + 0.2^2) = 0.36 / 0.40 = 0.9 . Read the observed colors carefully.

This question is coming soon to HackTheRounds.

Representative Question 6: Three-Dice Expected Value

Problem: Roll three fair six-sided dice. If all three match, win $20. If exactly two match, win $10. If all three differ, lose $2. Compute expected return per roll in dollars.

Compute each probability. All three match: 6 (1/6)^3 = 1/36 . All three differ: 6 5 4 / 6^3 = 120/216 = 5/9 . Exactly two match: 1 - 1/36 - 5/9 = 1 - 1/36 - 20/36 = 15/36 . Expected value: (1/36) 20 + (15/36) 10 + (20/36) (-2) = (20 + 150 - 40) / 36 = 130 / 36 ≈ 3.61 . Round to cents: $3.61 .

Time budget: two minutes if you are fluent with complementary counting. The trap is forgetting that "exactly two match" is not just C(3, 2) (1/6) (5/6) , which is wrong.

This question is coming soon to HackTheRounds.

Strategy Across the 17 Questions

The hardest meta-skill is time management. I classified every question into "under 30 seconds," "1 to 2 minutes," or "skip for now" within the first ten seconds of reading. Anything that went into "skip for now" I came back to only after finishing everything else. Five skipped questions returned eleven minutes of cleanup time in the last stretch, which is where I made up ground on two of the balance-system problems I could not immediately set up.

Result

Pending. SIG does not disclose scores or a passing cutoff. Recruiters say results arrive within ten business days.

Tips

  1. Practice mental arithmetic on fractions, not decimals. SIG's probability problems have clean fraction answers. Working in decimals loses you time on the conversion. Keep `1/36`, `5/9`, `15/36` in fractional form until the final conversion step.
  2. Write down the linear-system setup before you start substituting. The barn legs problem and the balance system problem are both systems-of-equations. Writing the equations on paper adds fifteen seconds and prevents a two-minute arithmetic error. SIG's grader does not give partial credit.
  3. Use complementary counting on dice problems. "Exactly two match" is `1 - P(all match) - P(all differ)`. Computing it directly via `C(3, 2) * p * q` gets the wrong answer because the "third die differs" constraint is easy to miscount. The complementary form is shorter and less error-prone.
  4. Flag-and-skip aggressively in the first pass. If a question takes longer than 90 seconds to set up, flag it and move on. Coming back with three minutes left on a fresh head is strictly better than grinding for four minutes and still missing it.
  5. Paper and pen next to your keyboard. SIG's proctoring setup does not provide built-in scratch space. The browser console works in a pinch, but paper is faster. Verify your proctoring rules beforehand; some sessions prohibit paper.
  6. Drill the six highest-frequency problem types the week before. Those are: barn-legs linear systems, round-table logic, age-bucket conditional probability, balance-system equations, Bayesian factory updates, and dice expected-value. Every SIG Quant OA in the last three cycles has used at least five of these six types. If you can solve all six from memory under two minutes each, you will finish with time to spare.

SIG's Quant OA is less about knowledge and more about steady pace under pressure. Seventeen questions in sixty minutes allows no rescue for a question that went sideways. The candidates who clear are the ones who defaulted to clean frameworks (indicator variables, complementary counting, linear systems) instead of improvising.