SIG Quant Researcher New Grad Interviews: 49 Experiences
- Reports
- 49 interview reports
- Level
- New Grad
- Stages
- OA → Phone screen → Onsite
SIG PhD quant researcher new grad interviews since late 2023: an online test, then a recruiter or HR call with quick probability questions.
At a glance
- Process: an online test (11 of the 20 current PhD reports), then a recruiter or HR call (14 of 20) that often mixed fit questions with two or three probability questions, then phone rounds, a data task and a final. Candidates 13, 15, 17 and 19 were cut right after the HR or recruiter round.
- Asked recently: probability, dice or coin questions in 17 of 20. The coin-flip pattern race came up in 5 reports (03, 15, 16, 17, 20), the bus-waiting question and the two-envelopes switch in 3 each, and "why SIG" or other motivation questions in 6.
- Online test: 17 questions in 60 minutes in the 2024 tests (08, 10) or 9 in an hour (12, 14), mostly probability and word puzzles; candidates said many had been shared online with new numbers.
- Practise first: which coin pattern shows up first, catching the bus to the movie, two envelopes with a known range and one number is the average of the other two.
Coverage: 49 SIG new grad Quantitative Researcher reports in one numbered sequence, each candidate counted once:
- PhD candidates, Dec 2023 to 2026 (01–20): current, and the base for the counts and tables below.
- Master's and bachelor's candidates (21–24): their own role section.
- History, 2014 to Oct 2023 (25–49): PhD reports, plus one master's candidate (28), outside the current counts.
PhDs who applied while already working are in SIG PhD Quantitative Researcher Experienced Hire OA / Interviews. Internship reports are in SIG PhD Quantitative Research Intern OA / Recruiter Call / Interviews.
What came up
These counts cover the 20 current PhD reports (01–20). Probability carried almost every first round; the recruiter call was a real filter.
| Topic | Reports | Last seen |
|---|---|---|
| Probability, dice and coin questions | 17 of 20 | Feb 2026 |
| Recruiter or HR call | 14 of 20 | Feb 2026 |
| Online test | 11 of 20 | Feb 2026 |
| Why SIG, motivation and fit | 6 of 20 | Dec 2025 |
| Algorithms and programming | 3 of 20 | Feb 2026 |
| Statistics, regression and Bayes | 2 of 20 | Feb 2026 |
| Data task or data round | 1 of 20 | Feb 2026 |
Rows: probability, dice and coin questions 01, 02, 03, 04, 06, 08, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20; recruiter or hr call 01, 02, 03, 05, 06, 08, 09, 13, 14, 15, 16, 17, 18, 19; online test 01, 02, 03, 06, 08, 09, 10, 11, 12, 14, 20; why sig, motivation and fit 02, 05, 06, 13, 17, 20; algorithms and programming 01, 04, 06; statistics, regression and bayes 01, 06; data task or data round 01.
Questions we found
These tables cover the 20 current PhD reports. "From a reply" marks questions another poster added under report 02.
Online test puzzles (unlinked): asked more than once
| Question | Reports | Last seen |
|---|---|---|
| A canoe or rowing trip: work out the start time | 2: 08, 10 | Nov 2024 |
| A balanced hanging mobile (96 lb): weight of one shape | 2: 10, 11 | Sep 2024 |
| An ages puzzle (who tells the truth to whom) | 2: 12, 14 | Jun 2024 |
| Shortest time across grass or a swamp and then a road | 2: 12, 14 | Jun 2024 |
| Two people arriving at random times: chance they meet | 2: 12, 14 | Jun 2024 |
| Expected flips until two heads in a row | 2: 12, 14 | Jun 2024 |
| Three biased coins or machines: a posterior probability | 2: 12, 14 | Jun 2024 |
| Chance the smallest or largest of three uniform numbers lands in a range | 2: 12, 14 | Jun 2024 |
Online test puzzles (unlinked): asked once
| Question | Reports | Last seen |
|---|---|---|
| Legs in a barn or cage (520 legs): how many of one animal? | 1: 10 | Sep 2024 |
| Five children or toddlers at a round table, with seating rules | 1: 10 | Sep 2024 |
| Read a preferences or profit table and find a probability | 1: 10 | Sep 2024 |
| Fake news: share of fake posts with likes as evidence | 1: 10 | Sep 2024 |
| Three dice paying $20 for a triple and $10 for a pair: expected return | 1: 11 | Aug 2024 |
| A small grid with black cells: placements or chances | 1: 12 | Jun 2024 |
| A path or duck moving with alternating step sizes | 1: 14 | May 2024 |
| Numbered blocks: a counting question | 1: 14 | May 2024 |
| A many-sided die: rolls divisible by 4, or an expectation | 1: 14 | May 2024 |
Probability and counting: asked more than once
| Question | Reports | Last seen |
|---|---|---|
| Coin flips: HTH before HHT | 5: 03, 15, 16, 17, 20 | May 2025 |
| Bob and a bus to a movie | 3: 03, 13, 20 | May 2025 |
| Two envelopes, one holds twice the other, the smaller amount is uniform on 0 to 100: when should you switch? | 3: 15, 16, 17 | Apr 2024 |
| Three numbers picked from 1 to 20: chance one is the average of the other two | 2: 13, 19 | May 2024 |
Probability and counting: asked once
Statistics, linear algebra and numerics
| Question | Reports | Last seen |
|---|---|---|
| Derive the OLS formula, then say when it is efficient | 1: 01 | Feb 2026 |
Logic puzzles
| Question | Reports | Last seen |
|---|---|---|
| An election with three candidates | 2: 01, 04 | Feb 2026 |
Algorithms and programming
| Question | Reports | Last seen |
|---|---|---|
| How to write a sorting algorithm, and its complexity (no code) | 1: 01 | Feb 2026 |
| Median of n numbers that are all between 1 and 100 (describe the method) | 1: 04 | Apr 2025 |
Data task
| Question | Reports | Last seen |
|---|---|---|
| A trading ledger for one asset between seven traders: propose a strategy that maximizes return | 1: 01 | Feb 2026 |
Behavioral (BQ)
| Question | Reports | Last seen |
|---|---|---|
| Why SIG? (from a reply) | 4: 02, 13, 17, 20 | Dec 2025 |
| Why quant research? Why finance? Research or trading? | 2: 13, 16 | May 2024 |
| What matters most to you: team, learning, pay or achievement? What matters most in a job? | 2: 16, 20 | Mar 2024 |
| Research or trading: which do you prefer? (from a reply) | 1: 02 | Dec 2025 |
| What you like to do (hobbies) | 1: 03 | May 2025 |
| Background, interests and motivation | 1: 05 | Apr 2025 |
| Motivation questions | 1: 06 | Dec 2024 |
| How did you first hear about SIG? | 1: 13 | May 2024 |
| Do you play Texas hold'em or other strategy games? Hobbies | 1: 16 | Mar 2024 |
| How does your research relate to quant work? | 1: 16 | Mar 2024 |
| A hard moment in your research; a time you showed leadership | 1: 20 | Dec 2023 |
20 PhD new grad reports, Dec 2023 to 2026
01: Economics PhD, out after the data task review
Report date: Feb 13, 2026 · Role: Quant Research & Systematic Trading (full-time, new grad) · Experience: economics PhD student · Result: No offer
Online assessment (November 2025)
- 17 questions, which the candidate posted separately.
First recruiter round (late December 2025)
- Mostly standard probability questions from the "green book," a popular quant interview prep book.
- A question the candidate said had already been shared online, about nine-sided dice and making a straight or a full house. It sounds like the five-dice reroll question intern candidates reported in 2025, though no numbers are given.
Quant phone screen (early January 2026)
- Derive the OLS formula (ordinary least squares, the standard way to fit a straight line to data), then find the condition that makes it efficient.
- A question about an election with three candidates.
- The candidate had asked the recruiter whether there would be algorithm or coding questions and was told no. The quant still asked how to write a sorting algorithm and what its complexity is, without writing code.
Data task (late January 2026, 3 days)
- A trading ledger for one asset, with trades between seven traders. The task: propose a strategy that maximizes return. Any programming language.
Data task review (virtual, two days before the post)
- Two quants, one hour each. One talked through the data; the other asked the candidate to present the strategy.
- The candidate came away thinking the strategy was weak. In hindsight, they felt the ledger should be read as traders reacting to each other, not mined for patterns in the price alone.
The recruiter had said an onsite would follow if this round went well.
02: PhD candidate invited to a 45-minute recruiter interview
Report date: Dec 10, 2025 · Role: Quant Research & Systematic Trading (full-time, new grad) · Experience: PhD student · Result: Not reported
- After a long wait following the online test, the candidate was invited to a 45-minute recruiter interview.
- The recruiter said it would include two or three math questions.
The candidate never posted how it went. A reply in February 2026, from someone who doesn't say they took the call, listed what the recruiter asks: whether you prefer research or trading, why SIG, a green-book-style probability question about waiting for a bus, and a simple Markov chain dice question (a Markov chain is a process where the next step depends only on where you are now). Two other repliers had just been invited too.
Practice: Research or trading: which do you prefer?, Why SIG?.
03: Math PhD, two probability questions on a recruiter call
Report date: May 5, 2025 · Role: Quant Research & Quant Trading (full-time, new grad) · Experience: math PhD, no quant internship · Result: Rejected about a week later
Online assessment: 17 questions in 60 minutes, most already known.
Recruiter phone interview (booked one to two weeks later, billed as 30 to 45 minutes with two or three math questions). It opened with what the candidate likes to do, then:
- Bob and a bus to a 4 pm movie: both reach the stop at random times between 3 and 4, the bus always stays 10 minutes, and the ride takes 20. The candidate answered 5/24 and the recruiter accepted it.
- Keep flipping a coin: chance HTH appears before HHT. The candidate called it a classic Markov chain question, with answer 1/3, and noted other ways to solve it.
Readers first got about 0.32 for the bus question. The candidate then explained the rule they had used: the bus always stays the full 10 minutes, whether or not Bob gets on, so it must arrive by 3:30, and the good region has area 750 out of 3600, which is 5/24.
Then the candidate's own questions. Both answers came quickly. On the call, the candidate was told the decision weighs the interview, the online test and the résumé together; they suspect the test (three wrong) or the résumé.
04: PhD, a phone screen with a median, dice and a vote
Report date: Apr 23, 2025 · Role: quant (title not given; full-time, new grad) · Experience: PhD student · Result: Rejected
Phone screen:
- Given n numbers that are all between 1 and 100, describe how to find the median.
- Roll 10 dice: chance the sum is divisible by 7.
- An election with three candidates (named only).
The candidate answered nearly everything and was still rejected.
05: Biostatistics PhD, an HR call about background and motivation
Report date: Apr 6, 2025 · Role: Quantitative Researcher (full-time, new grad) · Experience: PhD student in biostatistics (mostly theory), undergraduate math and computer science · Result: Not reported
- HR scheduled a phone interview to look at the candidate's background, interests and motivation.
- With no finance background, the candidate wasn't sure what SIG wanted to hear. The post never said how the call went.
06: Statistics PhD, one Bayes question on the HR call
Report date: Dec 19, 2024 · Role: Quantitative Researcher (full-time, new grad), Ireland · Experience: statistics PhD student · Result: Rejected two days later
- Online assessment: 18 questions, mostly probability and Bayes, most already known.
- HR call (about a week later): many motivation questions, then one Bayes question: a test comes back positive, so what is the true chance of having the disease? Follow-up after follow-up on the same question. No numbers were shared.
The candidate described SIG's HR call as very predictable: behavioral and motivation questions plus probability and statistics, with no coding, and the questions people report repeat a lot, mostly Bayes' rule, Markov chains and the law of total probability. They felt their answers were right and guessed the motivation part cost them. Another candidate for the same role replied two months later: same interviewer, similar content, and they moved on. Their take: the HR round also looks at how clearly you explain your working without writing it down, and asks you to explain a result by intuition, for example why a conditional chance differs so much from the plain one.
07: PhD, past the phone screen and still waiting
Report date: Dec 6, 2024 · Role: Quantitative Researcher (full-time, new grad) · Experience: PhD student · Result: Passed the phone screen, then still waiting
- The candidate was still in the phone-interview stage and worried the openings might already be filled.
In the replies, another candidate said they were turned down after their final video round that same week.
08: Final-year physics PhD, a 17-question test from a recruiter's email
Interview date: Nov 29, 2024 · Role: Quantitative Researcher (full-time) · Experience: final-year physics PhD student, one earlier internship outside finance · Location: U.S. (office not stated) · Result: Not reported
Recruiter step (email, no call)
- A recruiter on LinkedIn passed their CV to a SIG recruiter. That first recruiter had asked about an internship, but they only wanted full-time roles.
- The SIG recruiter emailed them a link to the online test. No call is described.
Online assessment: the Susquehanna Problem Solving Assessment (1 hour)
The test had 17 questions, and candidates weren't expected to finish all of them. They wrote down ten afterwards:
- Two factories: 40% of A's widgets are red, and 80% of B's are. One factory is chosen at random, two of its widgets are drawn, and both turn out red. How likely is it that they came from A? Practice: which of two factories made two red widgets.
- Two rolls of a fair die. Different numbers pay the larger one in dollars; a double pays nothing. What do you expect to win? Practice: paid the higher of two dice only if they differ.
- In a bakery, each customer buys one item: a croissant 70% of the time, a muffin 30% of the time. Two muffins are left and five people are in line. What's the chance no one wants a muffin after they run out? Practice: two muffins for five customers.
- You have 6 identical cookies of one kind and 8 identical cookies of another. How many ways can you put 7 of them in a row? Practice: lining up cookies from two kinds.
- Three cats enter a contest and win with chances 3/4, 3/16 and 1/16. You cheer for one cat picked at random, and it loses. What's the chance you picked the strongest cat? Practice: the cat you picked lost.
- A spinner has three regions with chances 1/4, 1/4 and 1/2. How many spins do you expect before it has landed in two different regions? Practice: spins until two different regions.
- Two flowers each open at a random moment in the next 20 days, independently. One stays open for 4 days, the other for 8. How likely is it that their blooms overlap at all? Practice: two flowers in bloom together.
- A deck has 8 cards, two each of 10, J, Q and K. You're dealt 5. How many pairs do you expect in your hand? Practice: pairs in a 5-card hand from an 8-card deck.
- You have 3 tokens and want to reach 5 before you run out. You play boldly: each turn you bet as much as you can, but never more than you need to reach 5. You win each bet with chance 3/4. What's the chance you reach 5? Practice: growing 3 tokens to 5 with bold betting.
- A frog goes from (0,0) to (5,4), one step up or right at a time, and never takes three steps in a row in the same direction. How many paths are there? Practice: frog paths with no three steps in a row in one direction.
The frog question was the last one. Before the test, a chatbot had suggested it to them as a sample SIG question, and they spent about 30 minutes on it. In the test they only got as far as the count without the rule (126 paths) and ran out of time. Afterwards they solved it by splitting each direction's steps into runs of one or two and matching the runs up, got 45, and checked that with a brute-force script.
Their takeaway was that some questions are harder than they look, so the easy ones are worth doing first.
09: PhD, a math assessment booked by HR
Report date: Oct 14, 2024 · Role: Quant Researcher (full-time, new grad) · Experience: PhD student · Result: Passed; a call with a quant came next
- Online assessment: many questions already known; it went well.
- Math assessment phone round: HR booked it about a week and a half later, billed as 30 to 45 minutes. The candidate passed and was scheduled with a quant. No questions were shared.
10: Sep 24, 2024 · PhD new grad · online test, then a rejection two weeks later
17 questions in 60 minutes. They thought most answers were right and guess they were put on hold. The test had legs in a barn, five toddlers at a round table, a bookstore profit table, a hanging mobile of 96 lb, fake news (30% fake, likes as evidence), the higher of two dice, enough muffins, who has the most cats, a canoe trip, the cat that lost, two flowers in bloom, a grid of 1 to 4 floor buildings, pairs in 5 cards from 8, bold betting from 3 tokens to 5 and frog paths to (5, 4).
11: Aug 19, 2024 · PhD new grad · online test
Their screenshots show two questions from the same 17-question test: three dice paying $20 for a triple, $10 for a pair and −$2 otherwise (the expected return), and the 96 lb hanging mobile.
12: Jun 22, 2024 · PhD new grad, through a headhunter · passed the online test
9 questions in 60 minutes, all shared online before with new numbers or names: a dog crossing a swamp and a road to a house (shortest time), three machines with different win chances (a conditional probability), an ages puzzle where people tell the truth only to older people, the expected number of wins from dice, two people who each wait a while (chance they meet), paths from (0, 0) to (5, 5) with steps of 1 and 2, the expected flips until two in a row, the chance the smallest of three numbers on 0 to 6 falls in [2, 3], and placing items on a 3×3 grid with 2 black cells. A second round was booked for the next day.
13: May 30, 2024 · PhD new grad · out after an 80-minute HR round
40 minutes on research, background and motivation: why SIG, why quant, how they first heard of SIG, trading or research. Then 40 minutes of probability with the recruiter coaching: a waiting-for-the-bus area question like the one on the test, and 20 balls, pick 3, chance one is the average of the other two. Their parity approach got messy; the recruiter hinted at symmetry.
Practice: Waiting for a bus with a deadline, or two people arriving at random times, Why SIG?, Why quant research?, Why finance?, Research or trading?, How did you first hear about SIG?.
14: May 6, 2024 · PhD new grad, no finance background · online test
They applied online, got an HR email an hour later and did the test that night: 9 questions in an hour. Their notes on each type: a path that alternates step sizes, a grass-and-road shortest time (bound it, since only seconds matter), an ages puzzle (read which person is asked about), a posterior over three choices, a many-sided die expectation, the chance an extreme of uniform numbers lands in a range, two people waiting 60 minutes (area of a square), numbered blocks (counting), and the expected flips to the first HH (set up two states).
15: Apr 7, 2024 · PhD new grad, graduate QR · out after the HR round
A 9-question, 60-minute test, then a 45-minute HR phone round with two questions: HHT before HTH and the two-envelope switch with the smaller amount uniform on 0 to 100 (find the best threshold). Both answers were right but the follow-ups took time. A rejection came three days later.
16: Mar 11, 2024 · PhD new grad, through a headhunter · passed the HR round
Soon after the test came an HR round: their research and how it relates to quant work, and what matters most in a job (they said colleagues). They brought up SIG's game culture themselves (poker; they prefer mahjong). Then two questions shared online before: a coin-flip question and choosing or switching envelopes. They asked about training for new hires. They passed two days later to a quant round.
Practice: Coin flips: HTH before HHT, HHT before HTT, Two envelopes, one holds twice the other, the smaller amount is uniform on 0 to 100: when should you switch?, What matters most to you: team, learning, pay or achievement?, Why quant research?, Why finance?, Research or trading?, How does your research relate to quant work?.
17: Feb 22, 2024 · PhD new grad, through a headhunter · out after the HR round
A 45-minute HR video round: 10 minutes of self-intro, PhD research and why SIG, then HTH before HHT (1/3) and the two-envelope switch on 0 to 100. The recruiter steered them to "never switch above 100, switch below it". A rejection came a week later. Another candidate in Sydney got the same two questions.
Practice: Two envelopes, one holds twice the other, the smaller amount is uniform on 0 to 100: when should you switch?, Why SIG?.
18: Feb 22, 2024 · PhD new grad · out after the first round
Two probability questions: 3 people sit at random at a round table with 8 seats (chance no two sit together), and 2 red and 2 blue balls drawn one at a time where you guess each color and win $1 per right guess (what would you pay to play?). A reader asked whether this was the HR math round.
Practice: A round table with 8 seats, 3 people sit at random: chance no two sit together, 2 red and 2 blue balls drawn one at a time, you guess each color and win $1 per right guess: what would you pay?.
19: Jan 17, 2024 · PhD new grad · out after the HR math round
The interviewer agreed with every answer: three numbers from 1 to 20, one the average of the other two, and an urn with 3 balls, 2 of them red, where each drawn ball is replaced by a blue one (expected draws until no red is left, 4.5). They were still rejected.
20: Dec 21, 2023 · PhD new grad · four rounds, then out
- Online test with probability and a Markov chain.
- Phone: behavioral questions, then the bus, an obtuse triangle from three random points on a circle and HHT before HTH.
- Resume and behavioral: research, markets, why SIG, which matters most (team, learning, pay or achievement), a hard moment in research, leadership.
- Three questions: a random black-and-white grid laid over its 90° rotation (chance it's all black), whether correlations 0.1, 0.8 and 0.7 can go together, and a recursion like the airplane seat puzzle.
A rejection came three working days later.
Practice: Three variables with corr(X, Y) = corr(Y, Z) = 0.9: what can corr(X, Z) be? Is 0.5 possible? Are 0.1, 0.8 and 0.7 possible together?, Each cell of a grid is black with chance 1/2; lay it over its 90° rotation: chance everything is black, Why SIG?, What matters most to you: team, learning, pay or achievement?, A hard moment in your research, a time you showed leadership.
Master's and bachelor's candidates
Four current reports from candidates without a PhD: three online tests (22, 23, 24) and one second round of three probability questions (21). The tests reused many of the PhD tests' puzzles.
Online test puzzles (unlinked)
| Question | Reports | Last seen |
|---|---|---|
| A path or duck moving with alternating step sizes | 2: 22, 24 | Sep 2024 |
| Legs in a barn or cage (520 legs): how many of one animal? | 1: 23 | Aug 2024 |
| Five children or toddlers at a round table, with seating rules | 1: 23 | Aug 2024 |
| Read a preferences or profit table and find a probability | 1: 23 | Aug 2024 |
| A balanced hanging mobile (96 lb): weight of one shape | 1: 23 | Aug 2024 |
| Fake news: share of fake posts with likes as evidence | 1: 23 | Aug 2024 |
| Three dice paying $20 for a triple and $10 for a pair: expected return | 1: 23 | Aug 2024 |
| An ages puzzle (who tells the truth to whom) | 1: 24 | Jun 2024 |
| Shortest time across grass or a swamp and then a road | 1: 24 | Jun 2024 |
| Two people arriving at random times: chance they meet | 1: 24 | Jun 2024 |
| Expected flips until two heads in a row | 1: 24 | Jun 2024 |
| Numbered blocks: a counting question | 1: 24 | Jun 2024 |
| Three biased coins or machines: a posterior probability | 1: 24 | Jun 2024 |
| Chance the smallest or largest of three uniform numbers lands in a range | 1: 24 | Jun 2024 |
| A many-sided die: rolls divisible by 4, or an expectation | 1: 24 | Jun 2024 |
21: Master's, a second round of three probability questions
Report date: Apr 30, 2026 · Role: Quantitative Researcher (the post's tags say full-time new grad; the post itself calls it an internship) · Experience: master's · Result: Not reported
- Online assessment: passed.
- Next round: a one-hour Zoom interview with three probability questions. The candidate asked what to expect.
One reply came from someone who remembered the round: brain-teaser-style probability questions, with the first round's a bit easier.
22: Sep 5, 2024 · bachelor's, full-time · online test
The 17-question test, with a new question: a duck that moves only up or right toward a point without overshooting and alternates step sizes.
23: Aug 26, 2024 · master's new grad · online test
Legs in a barn (520 legs), five toddlers at a table, a bookstore profit table, the 96 lb mobile, fake news (30% fake), three dice paying $20 or $10, and enough muffins (70% croissants).
24: Jun 13, 2024 · master's new grad, from a career fair · online test, not passed
9 questions: rolls of a 12-sided die divisible by 4, blocks numbered 1 to 6 plus two 7s, an ages puzzle, the expected flips to the first HH, a sheep crossing grass and a road, three biased coins (a posterior), the largest of three numbers on 0 to 5 in a range, a duck alternating steps to (13, 8), and a picnic meeting at random times.
History (2014–Oct 2023)
These 25 reports are more than three years old, so they stay out of the counts above. Report 28 is a master's candidate; the rest are PhDs. Their named questions:
Online test puzzles (unlinked)
| Question | Reports | Last seen |
|---|---|---|
| Three dice paying $20 for a triple and $10 for a pair: expected return | 2: 26, 41 | Oct 2023 |
Probability and counting: asked more than once
Probability and counting: asked once
| Question | Reports | Last seen |
|---|---|---|
| Waiting for a bus with a deadline, or two people arriving at random times | 1: 25 | Oct 2023 |
| Three random points on a circle: an obtuse triangle, or a triangle that holds the center | 1: 25 | Oct 2023 |
| 10 people at a round table, 3 are friends: chance at least two of them sit together | 1: 25 | Oct 2023 |
| A family has two children and one is a girl: chance the other is a girl | 1: 25 | Oct 2023 |
| A friend doubles the pot: the lowest chance of winning at which you should call (1/4) | 1: 26 | Oct 2023 |
| 99 fair coins and one with two heads, pick one and get 7 heads in 7 flips: chance it's the two-headed coin (128/227) | 1: 26 | Oct 2023 |
| Two players take turns rolling a die until a 6; expected rolls, and the same given the first player wins | 1: 27 | Aug 2023 |
| Roll 12 dice: chance the sum is under 30 (normal approximation, no calculator) | 1: 29 | Jun 2023 |
| Expected distance between two random points on a circle (4/π for a unit circle) | 1: 31 | Oct 2022 |
| 16 flips of a coin with heads chance p: expected number of runs of heads | 1: 33 | Sep 2022 |
| Easy HR questions: Bayes, drawing balls, coins, find the light ball among 8, four people crossing a river | 1: 34 | Mar 2022 |
| A 2 m rope tied to a wall, cut at a random point and keep cutting the wall piece until it's under 1 m: expected cuts (1 + ln 2) | 1: 35 | Nov 2021 |
| Random chords in a circle: expected number of regions | 1: 38 | Nov 2020 |
| A contract on a stock with a few possible prices: find the missing price, then the value | 1: 40 | Aug 2020 |
| A stock at $100 will be $50 or $200 tomorrow, and a casino takes even-money bets: make money | 1: 41 | Aug 2019 |
| A die's expected value, and the expected maximum of two dice | 1: 41 | Aug 2019 |
| Three cards with sides 1-2, 2-3 and 3-4; you and the interviewer each see one side: who can name the other side first? | 1: 41 | Aug 2019 |
| Cubes painted on some faces: a conditional probability | 1: 43 | Dec 2018 |
| A hopping-rabbit (stairs) variant from the green book | 1: 43 | Dec 2018 |
| 2 red and 2 blue balls drawn one at a time, you guess each color and win $1 per right guess: what would you pay? | 1: 45 | Jan 2018 |
Statistics, linear algebra and numerics
| Question | Reports | Last seen |
|---|---|---|
| A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank | 4: 38, 40, 41, 44 | Nov 2020 |
| A finite-difference derivative: the best step size given machine precision | 2: 40, 43 | Aug 2020 |
| Does A independent of B and of C mean A is independent of B and C together? | 1: 38 | Nov 2020 |
| Split money across three stocks whose volatilities are in ratio 1:2:3 | 1: 40 | Aug 2020 |
| Eigenvalues of a matrix with given diagonals | 1: 41 | Aug 2019 |
| Fit a regression when points are denser in some areas (weighted least squares) | 1: 41 | Aug 2019 |
| One sample from N(0, σ²): estimate the variance and give a confidence interval | 1: 41 | Aug 2019 |
| An n×n matrix with rank under n/2; M·Mᵀ = M: what follows? | 1: 42 | Jun 2019 |
Algorithms and programming
| Question | Reports | Last seen |
|---|---|---|
| Python basics; how many lines of code have you written; structs | 2: 35, 40 | Nov 2021 |
| Design a stock order book: data structures for each operation and their cost (the "system design" round) | 1: 37 | Mar 2021 |
| Shuffle a list at random; find the max and min across two sequences | 1: 42 | Jun 2019 |
Behavioral (BQ)
| Question | Reports | Last seen |
|---|---|---|
| Why SIG? | 6: 25, 26, 32, 34, 47, 48 | Oct 2023 |
| Which other companies are you talking to, and how far along? | 3: 33, 34, 48 | Sep 2022 |
| Why quant research? Why finance? Research or trading? | 2: 25, 26 | Oct 2023 |
| Do you play Texas hold'em or other strategy games? Hobbies | 2: 26, 31 | Oct 2023 |
| Large data sets, online ML courses, coding experience | 1: 26 | Oct 2023 |
| Your strength, a weakness, your favorite book (seen on the video call) | 1: 32 | Oct 2022 |
| The US or the Dublin office? | 1: 32 | Oct 2022 |
| GPA, school, GRE score, how you heard of SIG, why you fit the firm and the role | 1: 49 | Aug 2014 |
25: Oct 29, 2023 · PhD new grad, by referral · recruiter call
30 minutes: background, why finance, why SIG, then two questions: two children, one is a girl, chance the other is too; and 10 people at a round table with 3 friends, chance at least two of the friends sit together. The recruiter's line was noisy, like a listening test. Another candidate replied that their first round was a bus you can wait 10 minutes for, with a 10-minute ride, arriving within the hour (draw a square), a triangle holding the center and HHT before HTH, and that the second round was background and career interests.
Practice: 10 people at a round table, 3 are friends: chance at least two of them sit together, Why SIG?, Why quant research?, Why finance?, Research or trading?.
26: Oct 19, 2023 · PhD new grad · out after the first round
The recruiter, based in Dublin, had a physics background. Two probability questions first: a friend doubles a $10 pot to $20, so what's the lowest chance of winning at which you call (1/4); and 99 fair coins plus one with two heads, pick one and get 7 heads in 7 flips, chance it's the two-headed coin (128/227). Then behavioral: the PhD, why SIG, why quant research, hobbies, large data sets, online ML courses, coding experience, questions for them. A rejection came three days later; they think it was the lack of coding experience.
Practice: A friend doubles the pot: the lowest chance of winning at which you should call (1/4), Why SIG?, Why quant research?, Why finance?, Research or trading?.
27: Aug 4, 2023 · PhD new grad · phone round
A dice race question about expected rolls and conditional probability, stated unclearly over a bad phone line, with no follow-up. Use a good line.
28: Math master's, five rounds from the test to a trader
Interview date: July 2023 · Report date: Oct 2023 · Role: Quant Researcher (full-time, new grad) · Location: Bala Cynwyd · Result: Rejected after the trader round
- Online assessment (1 hour): probability, statistics and brainteasers.
- Recruiter technical screen (30 minutes): probability questions and behavioral questions.
- HR screen (30 minutes): a deep dive into past work and projects.
- Quant researcher technical interview (1 hour): probability, then a deep dive into the candidate's earlier quant research.
- Trader technical interview (1 hour): projects, brainteasers, and questions on arbitrage and market making.
No specific questions were shared.
29: Jun 26, 2023 · PhD new grad · out at round 3
Online test in late May, then two phone rounds of statistics and probability like the test. Round 3 was with a quant researcher of over ten years at SIG, announced as calculus, linear algebra, statistics and probability for 45 to 60 minutes: a resume walk-through, then two "simple" questions with no calculator, a Bayes coin question and the chance 12 dice sum to under 30 (normal approximation). They got the variance of one die (35/12) wrong and were rejected.
Practice: Roll 12 dice: chance the sum is under 30 (normal approximation, no calculator).
30: Jun 4, 2023 · PhD, full-time QR through a headhunter, status not given · data exercise next
Probability rounds at a "high school" level, all known questions, then an HR background chat, then a data exercise to be sent out and discussed at the onsite.
31: Oct 29, 2022 · PhD new grad · phone round
Resume, then "do you play Texas hold'em or similar strategy games?", then three numbers from 1 to 20, one the average of the other two and the expected distance between two random points on a unit circle (4/π).
Practice: Expected distance between two random points on a circle (4/π for a unit circle).
32: Oct 17, 2022 · PhD new grad, Dublin office · passed the recruiter round
They applied to the US and Dublin at the same time; SIG lets you, then picks one so you don't interview twice. The recruiter video round had no technical questions: introduce yourself, why SIG, the US or Dublin, their questions (visas), a unique strength, a weakness, the favorite book on the shelf behind them, hobbies. The US HR round came next, said to have two probability questions.
Practice: Why SIG?, strength, weakness.
33: Sep 6, 2022 · PhD new grad · out after the first phone round
Behavioral first (research, other companies), then 30 minutes on two questions: three numbers from 1 to 20, one the average of the other two, and the expected number of runs of heads in 16 flips of a coin with heads chance p. They needed many hints on the second one and think that sank them.
Practice: 16 flips of a coin with heads chance p: expected number of runs of heads.
34: Mar 1, 2022 · PhD new grad · out after the HR call
After round 1, an HR call: why SIG, other interviews, then several easy questions (conditional probability, Bayes, drawing balls, coins, finding the light ball among 8, four people crossing a river). The recruiter agreed with every answer, yet a rejection came the next day. They think they didn't sound keen enough on SIG when asked about other companies.
Practice: Why SIG?.
35: Nov 26, 2021 · PhD new grad, by referral · out after the first phone round
Background, research, some Python basics, then puzzles. The first: a 2 m rope tied to a wall is cut at a random point; stop if the wall piece is under 1 m, otherwise keep cutting the wall piece (expected cuts: 1 + ln 2). The rest were questions shared online.
36: Oct 11, 2021 · PhD new grad · online test
The test reused many known questions: candy, derivatives, matrices, plus simple probability and math in the first section.
37: Mar 17, 2021 · PhD new grad · onsite
The onsite was billed as quantitative, technology and "system design" rounds. Someone who had it replied that the system design round meant designing a stock order book: choosing data structures for each operation and their time cost, not designing distributed systems.
38: Nov 9, 2020 · PhD new grad, through a headhunter · two rounds
Round 1: the expected number of regions when n random chords are drawn in a circle, HTH vs HHT, and the range of corr(Y, Z) when corr(X, Y) = corr(X, Z) = 0.9. Round 2, with a head of quant research: sampling uniformly from the triangle with corners (1, 0, 0), (0, 1, 0) and (0, 0, 1); the highest rank of an n×n matrix with A³ = 0; and whether A independent of B and of C means A is independent of both together.
Practice: Three variables with corr(X, Y) = corr(Y, Z) = 0.9: what can corr(X, Z) be? Is 0.5 possible? Are 0.1, 0.8 and 0.7 possible together?, Random chords in a circle: expected number of regions, A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank.
39: Final-year math PhD, five steps from OA to a virtual onsite
Interview date: Fall 2020 · Report date: Jan 26, 2021 · Role: Quantitative Researcher (full-time) · Experience: fifth-year math PhD student, graduating 2021 · Location: U.S. (virtual onsite) · Result: Offer; the candidate later joined SIG as a quantitative researcher
Recruiter step (email)
- The candidate emailed a SIG recruiter they had met at a campus event a year earlier.
- The recruiter sent them the online test.
Online assessment
- Questions on probability, statistics, calculus and linear algebra.
- The first phone interview invite came soon after.
Two phone interviews
- The questions got harder as the process went on.
- When the candidate got stuck, the interviewers pointed them in the right direction.
Data exercise
- Part of the process, with no details shared.
Full-day onsite (virtual in 2020)
- No details shared.
The candidate said they couldn't share specific questions. They also said the process was much like the one an earlier PhD intern from their department described, apart from the online test. That 2019 intern loop is entry 51 in the PhD internship roundup.
40: Aug 7, 2020 · PhD new grad · out after three phone rounds
- A contract on a stock with a few possible prices (first find a missing price from the $10 expected value), two children who decide whether to bring a coin (chance they go to the movie), and HHT before HTT.
- Python (lines of code written so far, structs), a sock drawer where two random socks are both red with chance 1/2, a coin whose chance of heads has distribution F(p) = p² (chance of 2 tails in 3), and the best step for a forward difference with machine precision 1e-16.
- The most 1s in an n×n 0/1 matrix with A² = 0, a uniform point on the triangle with corners (3, 0, 0), (0, 4, 0) and (0, 0, 2) (the distribution of x, and E[x] + E[y] + E[z]), and splitting money across three stocks with volatilities in ratio 1:2:3.
Practice: A contract on a stock with a few possible prices: find the missing price, then the value, Two children decide on their own whether to bring a coin to decide on a movie: chance they go, A coin with an unknown chance of heads drawn from a given distribution: chance of 2 tails in 3, or 1 head in 3, A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank.
41: Aug 21, 2019 · PhD new grad, by referral · eight-round onsite
Three phone rounds, then an onsite from 9am to 5pm with eight rounds (one interviewer couldn't be reached and was made up later). Their questions, order mixed: two people flip 3 coins each (chance of the same number of heads); three horses paying $2, $4 and $6 per $1 (best bets); Monte Carlo for π and e; a regression with points denser in some areas (weighted least squares); eigenvalues of a matrix with 1s and 2s on the diagonal and r next to it; a die's expected value and the expected max of two dice; three normal variables with known correlations, chance X + Y > Z + 2; the $100 stock that goes to $50 or $200 with a 1:1 casino bet; take a die's value or roll again up to twice, then up to 50 times (to 3 decimals); estimate σ² from one N(0, σ²) draw and give a confidence interval; 8 coins vs 10 coins, chance of more heads; two dice each, I win $1 if any of yours matches mine; the rank of A when A² has rank 0; and three cards with sides 1-2, 2-3 and 3-4, where each player sees one side and tries to name the other first.
Practice: You and the interviewer each flip 3 coins: equal head counts pay the interviewer $2, otherwise you get $1, Three horses pay 2, 4 and 6 times a $1 ticket: how do you bet? (there is a sure profit), Three normal variables with all correlations 0.5: chance X + Y > Z + 2, I have 8 coins, you have 10: chance I get more heads, Each player has two dice; I win $1 if any of your numbers matches one of mine. Do you play? (105/216), Roll a die and keep it, or pay to roll again: the expected payoff with one paid reroll, then with many, A stock at $100 will be $50 or $200 tomorrow, and a casino takes even-money bets: make money, uniform points in a disk: how many of 1000 tries land inside, and the spread, A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank.
42: Jun 19, 2019 · PhD new grad · out after round 3
Round 3 was with SIG's global head of quant: resume, then linear algebra (an n×n matrix with rank under n/2; M·Mᵀ = M), a horse-race bet with a guaranteed profit (not a plain probability question), and two short algorithm questions (shuffle at random; the max and min of two sequences). The interviewer said SIG has no fixed yearly headcount: strong candidates get offers.
Practice: Three horses pay 2, 4 and 6 times a $1 ticket: how do you bet? (there is a sure profit).
43: Dec 13, 2018 · PhD new grad · first two rounds, out at the last question
A green-book hopping-rabbit variant; is anything wrong with this correlation matrix; painted cubes (a conditional probability; the author later recalled that corner cubes have three painted faces, edge cubes two and centre cubes one, much like the coin question where some coins have heads on both sides); red and black socks, work back to the counts from a probability (more than one answer works); a coin whose heads chance p has a given distribution, chance of 1 head and 2 tails; and the best step size for a finite-difference first derivative. They failed the last one. Advice: accuracy matters more than speed.
Practice: Three variables with corr(X, Y) = corr(Y, Z) = 0.9: what can corr(X, Z) be? Is 0.5 possible? Are 0.1, 0.8 and 0.7 possible together?, A coin with an unknown chance of heads drawn from a given distribution: chance of 2 tails in 3, or 1 head in 3.
44: Oct 31, 2018 · PhD new grad, from a campus career fair · three phone rounds
Each round had a little resume and skills talk. Five questions: two dice each, I win $1 if any of your numbers matches one of mine (105/216 for me, so play); 3 coins each, equal heads pay $2 to me and different pay $1 to you; three horses paying 2, 4 and 6; generate uniform points in a disk, then the expected number and standard deviation of 1000 points landing inside; and the rank of A when A² = 0. They think the last one sank them. Asked later about the disk question, the author explained the method: draw x and y each uniform on [0, 1] and keep the point when x² + y² ≤ 1, so the question is how many of the 1000 tries are kept.
Practice: You and the interviewer each flip 3 coins: equal head counts pay the interviewer $2, otherwise you get $1, Three horses pay 2, 4 and 6 times a $1 ticket: how do you bet? (there is a sure profit), Each player has two dice; I win $1 if any of your numbers matches one of mine. Do you play? (105/216), uniform points in a disk: how many of 1000 tries land inside, and the spread, A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank.
45: Jan 12, 2018 · PhD new grad · out at the phone round
They were asked about a past internship, then: 100 coin flips, chance of more than 60 heads; the horse race; 2 red and 2 blue balls, guess the color and draw (chance of a right guess); and 3 coins each, chance of the same number of heads. The interviewer said they had about even odds to pass. They didn't.
Practice: You and the interviewer each flip 3 coins: equal head counts pay the interviewer $2, otherwise you get $1, Three horses pay 2, 4 and 6 times a $1 ticket: how do you bet? (there is a sure profit), 2 red and 2 blue balls drawn one at a time, you guess each color and win $1 per right guess: what would you pay?.
46: Mar 18, 2017 · PhD, full-time QR, status not given · passed round 2
Round 1: a 15-minute fit call with a recruiter. Round 2: background, one finance question, one probability question and one game theory question. Another candidate later replied that their interviewer described five rounds in all: team member, data exercise, then onsite.
47: Mar 17, 2017 · PhD new grad, changing fields · HR call with a quant
HR asked about the resume, why they were changing fields and why SIG. Then a quant joined and asked three questions: 8 coins against 10 coins (chance of more heads); a die where the first roll is free, you get its value, and each reroll costs $1 (once, then unlimited); and three normal variables with all correlations 0.5, chance X + Y > Z + 2. They felt it went badly.
Practice: Three normal variables with all correlations 0.5: chance X + Y > Z + 2, I have 8 coins, you have 10: chance I get more heads, Roll a die and keep it, or pay to roll again: the expected payoff with one paid reroll, then with many, Why SIG?.
48: Jan 18, 2017 · PhD new grad, QR analyst, through a headhunter · phone round
It opened with why SIG, then detailed questions about other offers and how far along those interviews were (both HR and the technical interviewer asked this in detail), then any finance experience. Three questions: 3 coins each with $2 and $1 payoffs (play or not); two children who decide on their own whether to bring a coin, which then decides if they go to the movie (the chance they go); and corr(x, y) = corr(y, z) = 0.9 with corr(x, z) = 0.5 (is anything wrong?).
Practice: Three variables with corr(X, Y) = corr(Y, Z) = 0.9: what can corr(X, Z) be? Is 0.5 possible? Are 0.1, 0.8 and 0.7 possible together?, You and the interviewer each flip 3 coins: equal head counts pay the interviewer $2, otherwise you get $1, Two children decide on their own whether to bring a coin to decide on a movie: chance they go, Why SIG?.
49: Aug 28, 2014 · PhD, quant associate · HR screen
GPA, school, GRE score, how they heard of SIG, why they fit the firm and the role. Someone was to reach out about the next round.
Practice: how you heard of SIG, why you fit the firm and the role.
Patterns across the reports
- The recruiter or HR call is a probability round. Fourteen of the 20 current PhD reports describe it, and in several it mixed fit questions with one to three probability questions (for example 03, 06, 13); four candidates were cut right after it (13, 15, 17, 19).
- The online test reuses puzzles. The same word puzzles (a hanging mobile, an ages puzzle, a shortest time across grass and a road, two people meeting at random times) appear across 2024 tests from PhD and non-PhD candidates (10, 11, 12, 14, 22–24).
- "Why SIG?" is the steady behavioral question. It came up in current reports 02, 13, 17 and 20 and in many older ones.
Common questions
How long is the SIG new grad online test?
In 2024 it was 17 questions in 60 minutes (08, 10) or 9 questions in an hour (12, 14), mostly probability and word puzzles.
Does the SIG recruiter call have probability questions?
Usually yes: fourteen current PhD reports describe the recruiter or HR call, and several got one to three probability questions on it, such as the bus-waiting and coin-pattern questions (03).
Does SIG reuse its online test questions?
Candidates said so (12), and the same puzzles appear across 2024 reports (see Patterns, point 2).
Keep practicing
- Practice from these reports: five-dice reroll question (01); Bob and a bus to a 4 pm movie (03, 20, 25); Keep flipping a coin: chance HTH appears before HHT (03, 15, 17, 20, 25, 38, 40); Given n numbers that are all between 1 and 100, describe how to find the median (04); Roll 10 dice: chance the sum is divisible by 7 (04); which of two factories made two red widgets (08); paid the higher of two dice only if they differ (08, 10); two muffins for five customers (08, 10, 23); lining up cookies from two kinds (08); the cat you picked lost (08, 10); spins until two different regions (08); two flowers in bloom together (08, 10); pairs in a 5-card hand from an 8-card deck (08, 10); growing 3 tokens to 5 with bold betting (08, 10); frog paths with no three steps in a row in one direction (08, 10); who has the most cats (10); 20 balls, pick 3, chance one is the average of the other two (13, 19, 31, 33); an obtuse triangle from three random points on a circle (20, 25); Two envelopes, one holds twice the other, the smaller amount is uniform on 0 to 100: when should you switch? (15, 16, 17); Three variables with corr(X, Y) = corr(Y, Z) = 0.9: what can corr(X, Z) be? Is 0.5 possible? Are 0.1, 0.8 and 0.7 possible together? (20, 38, 43, 48); You and the interviewer each flip 3 coins: equal head counts pay the interviewer $2, otherwise you get $1 (41, 44, 45, 48); Three horses pay 2, 4 and 6 times a $1 ticket: how do you bet? (there is a sure profit) (41, 42, 44, 45); Three normal variables with all correlations 0.5: chance X + Y > Z + 2 (41, 47); I have 8 coins, you have 10: chance I get more heads (41, 47); Each player has two dice; I win $1 if any of your numbers matches one of mine. Do you play? (105/216) (41, 44); Roll a die and keep it, or pay to roll again: the expected payoff with one paid reroll, then with many (41, 47); Roll 12 dice: chance the sum is under 30 (normal approximation, no calculator) (29); Two players take turns rolling a die until a 6; expected rolls, and the same given the first player wins (27); Expected distance between two random points on a circle (4/π for a unit circle) (31); 16 flips of a coin with heads chance p: expected number of runs of heads (33); An urn with 2 red and 1 blue ball, each drawn ball is replaced by a blue one: expected draws until no red is left (4.5) (19); 10 people at a round table, 3 are friends: chance at least two of them sit together (25); 2 red and 2 blue balls drawn one at a time, you guess each color and win $1 per right guess: what would you pay? (18, 45); A friend doubles the pot: the lowest chance of winning at which you should call (1/4) (26); Each cell of a grid is black with chance 1/2; lay it over its 90° rotation: chance everything is black (20); Random chords in a circle: expected number of regions (38); A contract on a stock with a few possible prices: find the missing price, then the value (40); Two children decide on their own whether to bring a coin to decide on a movie: chance they go (40, 48); A coin with an unknown chance of heads drawn from a given distribution: chance of 2 tails in 3, or 1 head in 3 (40, 43); A stock at $100 will be $50 or $200 tomorrow, and a casino takes even-money bets: make money (41); uniform points in a disk: how many of 1000 tries land inside, and the spread (41, 44); A square matrix with A² = 0 (or rank 0): its rank, or the most 1s it can have; A³ = 0: the highest rank (38, 40, 41, 44).
- Behavioral prompts from these reports: Research or trading: which do you prefer? (02); Why SIG? (02); What matters most to you: team, learning, pay or achievement? (16, 20); Why quant research? (13, 16, 25, 26); Why finance? (13, 16, 25, 26); strength (32); weakness (32); how you heard of SIG (49); why you fit the firm and the role (49); How does your research relate to quant work? (16); A hard moment in your research (20); a time you showed leadership (20).
- SIG interview questions: SIG's interview page with its practice questions.
- SIG practice problems: every SIG question in the bank, including the probability ones. Needs sign-in.
- SIG hiring insights: SIG's online assessment and interview insights.
- Interview experiences: the full Experience feed.
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.