SIG (Susquehanna) Interview Questions & Probability Prep
Susquehanna International Group (SIG) interviews test how you make decisions under uncertainty using probability, odds, expected value, and game theory. On its careers and decision science pages, SIG says it uses puzzles to emulate its trading environment and uses the betting card game Wizard to introduce candidates to gaming at the firm. You will work through conditional probability tables, expected value calculations, and cognitive bias traps that mirror how SIG traders price risk across options, stocks, and ETFs. Soreno, an AI voice interview practice app for finance and banking candidates, includes free paper LBO and valuation drills with no sign-up, and voice mock interviews that test your accounting, valuation, and fit answers out loud.
Updated
- Probability, odds, and expectancy. calculating expected value across discrete outcomes and deciding whether a bet or trade has positive expectation.
- Decision making under uncertainty. acting on imperfect information, observing opponents, and updating estimates when new evidence appears.
- Decision science and cognitive biases. avoiding confirmation bias, anchoring, availability bias (such as the gambler's fallacy), and framing traps when pricing securities.
- Pruning decision trees under time pressure. identifying the most important pieces of information in a position without trying to calculate every branch.
- Role-specific technical skills. options theory and market making for trading roles, quantitative modeling for research roles, and systems architecture or SQL, VBA, and Python for technology and operations roles.
Key facts, benchmarks, and worked examples
Below is a breakdown of sig interview questions, sourced from official firm, school, and regulatory documentation.
Why SIG interviews use poker, Wizard, and strategy games5 points
On its Game Theory and Decision Science page, SIG explains that it uses poker to teach new traders about decision making under uncertainty. When a trader evaluates the expected value of a trade and decides how to price risk, they go through the same thought process as a poker player sizing a bet with incomplete information. Susquehanna is also home to World Series of Poker Bracelet winners Bill Chen and Jerrod Ankenman, who co-authored The Mathematics of Poker. During recruiting and training, SIG uses several specific games to test and build trading instincts. You do not need to be a tournament player before you apply, but you do need to understand the math and psychology each game tests.
1. Wizard
Wizard
A trick-taking card game that includes bidding and betting. Wizard is easy to learn and hard to master, and the firm uses it to introduce job candidates to gaming at Susquehanna.
2. Poker
Poker
Used to teach expected value, odds, and risk pricing under uncertainty. Trading interns and instructors also play at the office poker tables during lunch and after work.
3. Chess
Chess
In chess, you cannot calculate every branch of a decision tree. SIG highlights how skilled players prune the decision tree under time pressure and pick out the most important features of a position to form a plan, just as traders do in fast markets.
4. Magic
Magic
The Gathering: Requires decision making with imperfect information, observing opponents, and maximizing your probability of winning. SIG notes that its head of trading in Sydney is a professional Magic: The Gathering player.
5. Strategic board games
Strategic board games and SET
SIG traders and interns regularly play Backgammon, Catan, Dominion, Hanabi, Power Grid, Love Letter, The Resistance: Avalon, and SET to practice pattern recognition and quick decisions.
SIG's four official decision science problems, worked step by step4 rows
Susquehanna publishes four interactive decision science problems to show how cognitive biases distort probability estimates and security pricing. Below is each problem from SIG's site, what it tests, and the exact math behind it.
SIG's Confirmation Bias dataset: Rain vs. Pain (29 total observations)
| Condition | Rain | No Rain | Row Total (By our calculation from SIG's table) |
|---|---|---|---|
| Pain | 14 | 6 | 20 |
| No Pain | 7 | 2 | 9 |
| Column Total | 21 | 8 | 29 |
| Conditional Probability of Pain | 14 / 21 = 2/3 (66.7%) | 6 / 8 = 3/4 (75.0%) | 20 / 29 (69.0%) |
Part 1
First is Confirmation Bias and conditional (1)
First is Confirmation Bias and conditional probability tables. SIG presents a two-by-two table tracking joint pain and rainy days across 29 observations: 14 days with Rain and Pain, 6 days with No Rain and Pain, 7 days with Rain and No Pain, and 2 days with No Rain and No Pain. SIG asks: what is the relationship between pain and rainy days?
Part 2
If you only look at the (2)
If you only look at the 14 Rain/Pain cases and overweight that single cell, you might claim a positive correlation. SIG points out that this is illusory correlation, a form of confirmation bias. To find the true relationship, compare conditional probabilities across both columns. On rainy days, 1 out of 21 people feel pain, which is 14/21 or 2/3 (about 66.7 percent). On days with no rain, 1 out of 8 people feel pain, which is 6/8 or 3/4 (75 percent). Because 2/3 is less than 3/4, there is a slight negative correlation between rain and pain. In trading, SIG notes that you must search for data that disconfirms your fair price estimate as well as data that supports it.
Part 3
Second is Anchoring and insufficient adjustment (3)
Second is Anchoring and insufficient adjustment. SIG asks you to write down the last two digits of your Social Security number, decide whether the average annual temperature in Chugwater, Wyoming is higher or lower than that two-digit number in degrees Fahrenheit, and then estimate Chugwater's average annual temperature. Although a Social Security number has nothing to do with Wyoming weather, people with higher two-digit numbers submit higher temperature estimates. In market making, SIG warns that overweighting an initial price estimate causes traders to make adjustments that are too small when new market information arrives.
4. Third is Availability
Third is Availability Bias and the gambler's fallacy. SIG asks
if you flip a fair coin 10 times and get 10 heads in a row, is the 11th flip more likely to be heads or tails? As SIG states, the initial probability of flipping 10 consecutive heads on a fair coin is 1 in 1,024. Because a 1-in-1,024 streak stands out in your memory, availability bias tempts you to think tails is due. Because the coin is fair and flips are independent, the 11th flip is still equally likely to be heads or tails.
Part 5
Fourth is Framing and variance versus (5)
Fourth is Framing and variance versus expected value. SIG cites Amos Tversky and Daniel Kahneman's 1981 Science study on a disease outbreak expected to kill 600 people. In the first frame, Program A saves 200 people for certain, while Program B has a 1/3 chance that all 600 are saved and a 2/3 chance that no one is saved. In the second frame, Program C results in 400 people dying for certain, while Program D has a 1/3 chance that no one dies and a 2/3 chance that 600 people die. Every program has the same expected value: 200 people expected to live and 400 expected to die (with Programs B and D having higher variability). Yet 72 percent of physicians in the study chose A over B, while 78 percent chose D over C. By our calculation from SIG's cited Tversky and Kahneman figures, that is a 50-percentage-point swing toward the higher-variance option (from 28 percent choosing B to 78 percent choosing D) caused by framing identical math in terms of losses instead of gains.
Worked examples: 4 probability and game theory archetypes9 points
On its Interns and Co-ops page, SIG explains that a trading role mostly involves making markets, which means posting buy prices (bids) and sell prices (asks) on options, stocks, and ETFs to provide liquidity. In technical interviews, you combine conditional probability, expected value, and game theory to price uncertain payouts. You can also see how we break down business math in our consulting case study examples to build structured speed before your round.
Part 1
Here are four original worked examples (1)
Here are four original worked examples written by our team covering the core mathematical archetypes tested in quantitative trading and research interviews. All scenarios and numbers are invented for practice.
2. Archetype 1
Archetype 1
Bayes' theorem and market-making price updates. You are making a market on a binary contract that pays $100 if a company beats earnings tonight and $0 if it misses. Your research model says the company has a 40 percent prior probability of beating earnings. Just before you quote a price, an institutional order flow signal flashes positive. Historically, when a company beats earnings, this signal flashes positive 75 percent of the time. When a company misses earnings, the signal flashes positive 25 percent of the time. What is the new fair value of the contract, and how should you update your price?
3. Step 1
Step 1
Define the prior probabilities. Let B be the event that the company beats earnings, so P(B) = 0.40, and let M be a miss, so P(M) = 0.60. Step 2: Define the likelihoods of the positive signal (S). P(S | B) = 0.75 and P(S | M) = 0.25. Step 3: Calculate the total probability of seeing a positive signal, P(S) = (0.40 * 0.75) + (0.60 * 0.25) = 0.30 + 0.15 = 0.45. Step 4: Apply Bayes' theorem to find the updated probability of a beat given the signal, P(B | S) = 0.30 / 0.45 = 2/3, or about 66.67 percent. Step 5: Compute the expected value and avoid anchoring. The expected payout of the $100 contract is now (2/3) * $100 + (1/3) * $0 = $66.67. Instead of anchoring near the initial $40.00 value, center your two-sided bid-ask spread around the updated $66.67 expectation, such as bidding $65 and offering $68.
4. Archetype 2
Archetype 2
Poker pot odds, expected value, and bluff-catching. You are playing heads-up poker on the river, the final betting round. The pot holds $100. Your opponent bets $50, so the pot now holds $150 and it costs you $50 to call. Your hand can only beat a bluff, and loses to any value hand. What is the minimum probability that your opponent is bluffing for your call to break even in expected value?
5. Step 1
Step 1
Identify your risk and reward. You risk $50 to win a total pot of $200 ($100 initial pot + $50 opponent bet + $50 your call). Step 2: Set up the expected value equation. Let p be the probability that your opponent is bluffing. If you call and they are bluffing, your net gain is $150. If you call and they have a value hand, which happens with probability (1 − p), your net loss is $50. Thus, EV(Call) = p * $150 − (1 − p) * $50. Step 3: Solve for the break-even threshold where EV(Call) = $0. Setting 150p − 50 + 50p = 0 gives 200p = 50, so p = 50 / 200 = 1/4, or 25 percent. If your opponent bluffs here more than 25 percent of the time, calling has positive expected value.
6. Archetype 3
Archetype 3
Coin-toss streak expectations with Markov states. What is the expected number of fair coin flips required to get two consecutive heads (HH)?
7. Step 1
Step 1
Define the states by how close you are to the goal. State 0 is the start (or whenever you just flipped Tails), with 0 consecutive heads. State 1 is when your most recent flip was a single Head. State 2 is the absorbing goal state (HH). Step 2: Write the linear expectation equations. Let E0 be the expected flips needed from State 0, and E1 be the expected flips needed from State 1. From State 0, you flip once: with probability 0.5 you get Tails and stay in State 0, and with probability 0.5 you get Heads and move to State 1, so E0 = 1 + 0.5 * E0 + 0.5 * E1. From State 1, you flip once: with probability 0.5 you get Heads and finish (0 flips left), and with probability 0.5 you get Tails and reset to State 0, so E1 = 1 + 0.5 * E0. Step 3: Substitute E1 into the first equation: E0 = 1 + 0.5 * E0 + 0.5 * (1 + 0.5 * E0) = 1.5 + 0.75 * E0. Subtracting 0.75 * E0 from both sides gives 0.25 * E0 = 1.5, so E0 = 6 flips.
8. Archetype 4
Archetype 4
Asymmetric coin-pattern races (HHT versus HTH). Player A waits for Heads-Heads-Tails (HHT) and Player B waits for Heads-Tails-Heads (HTH) on repeated flips of a fair coin. Why does HHT take an average of 8 flips to appear while HTH takes an average of 10 flips, even though every 3-flip sequence has a 1/8 probability on any three given flips?
9. Step 1
Step 1
Check what happens when each player misses on the third flip. If Player A reaches HH and flips H instead of T, Player A is still sitting on two consecutive heads (HH) and only needs a T on the next flip to win. If Player B reaches HT and flips T instead of H (giving HTT), that second T resets Player B all the way back to State 0. Step 2: Write the Markov state equations for HTH (State 0 = start, State 1 = H, State 2 = HT): E0 = 1 + 0.5 * E1 + 0.5 * E0, E1 = 1 + 0.5 * E1 + 0.5 * E2, and E2 = 1 + 0.5 * (0) + 0.5 * E0. Solving from the top down gives E1 = E0 − 2 and E2 = 1 + 0.5 * E0. Substituting both into the middle equation yields E0 − 2 = 1 + 0.5 * (E0 − 2) + 0.5 * (1 + 0.5 * E0), which simplifies to 0.25 * E0 = 2.5, so E0 = 10 flips versus 8 flips for HHT.
What interviewers listen for in behavioral and fit questions4 points
On its Who We Are page, SIG lists six core values that guide how employees work together: Good Decision Making, Teamwork, Belonging, Growth Mindset, Expertise, and Integrity. In behavioral conversations, interviewers test whether you treat decision making as a skill that requires reflection and practice, and whether you agree with SIG's stated view that smart people working together always come up with a better solution than the lone genius. You can rehearse your stories out loud on our behavioral interview practice page to make sure your answers are concise and specific.
1. Good Decision Making
Good Decision Making
Show a time you took an analytical approach to a choice under uncertainty, priced the risk, and reflected on your process afterward.
2. Teamwork and Belonging
Teamwork and Belonging
Share an example where working with teammates improved an idea, or where questioning an assumption helped the group grow.
3. Growth Mindset
Growth Mindset
Emphasize asking questions on the desk and learning from mentors even when starting with no prior finance experience.
4. Integrity
Integrity
Be honest about what you know and what you do not know during live technical questioning.
Practice a SIG-style decision science question out loud1 detail
Overview 1
Below is a practice prompt modeled (1)
Below is a practice prompt modeled on SIG's game theory and decision science criteria, complete with an example answer written by our team. Once you have read the example, test your own spoken structure and math under time pressure on Soreno, where you can Practice a case free, no sign-up.
From Susquehanna: Game Theory + Decision Science, Susquehanna: Trading Careers, Susquehanna: Interns + Co-ops, Susquehanna: Who We Are, Susquehanna: Careers Overview and Susquehanna: Technology Careers.
What SIG tests and how its recruiting programs work
Susquehanna operates at the intersection of trading, quantitative research, and technology across offices including Bala Cynwyd (Pennsylvania), New York, Dublin, and Sydney. Across its Quantitative Trading, Quantitative Research and Machine Learning, Technology, and Operations programs, the firm evaluates how you approach problems when information is incomplete and the clock is running. For students who are not yet eligible for its 10-week internship program, SIG also runs Discovery Programs in trading and technology that introduce students to the firm's environment and provide an early introduction to the internship interview process. On its trading careers page, SIG states that it uses puzzles during recruiting to emulate the trading environment and show candidates what it is like to work at the firm. Rather than checking whether you have memorized finance formulas, interviewers watch how you structure a problem, update your probability estimates as new data arrives, and size your risk. If you want to warm up on out-loud math and structured reasoning on Soreno right away, you can Practice a case free, no sign-up, or see a real sample report to check how your spoken math and logic are scored.
Probability, odds, and expectancy
calculating expected value across discrete outcomes and deciding whether a bet or trade has positive expectation.
Decision making under uncertainty
acting on imperfect information, observing opponents, and updating estimates when new evidence appears.
Decision science and cognitive biases
avoiding confirmation bias, anchoring, availability bias (such as the gambler's fallacy), and framing traps when pricing securities.
Pruning decision trees under time pressure
identifying the most important pieces of information in a position without trying to calculate every branch.
Role-specific technical skills
options theory and market making for trading roles, quantitative modeling for research roles, and systems architecture or SQL, VBA, and Python for technology and operations roles.
Official sources: Susquehanna: Game Theory + Decision Science, Susquehanna: Trading Careers, Susquehanna: Interns + Co-ops, Susquehanna: Who We Are, Susquehanna: Careers Overview and Susquehanna: Technology Careers.
1 practice question and sample answer
The practice prompts and sample answers below are Soreno's own, written for invented candidates.
Practice question and sample answer1 question
It states the backward-induction structure in the very first sentence, explicitly accounts for the $0.50 reroll cost before setting the threshold, notes the variance tie-breaker on rolling a 3, and walks through the numerator (24) and denominator (6) out loud so the interviewer can follow every step.
You are offered a game with a standard six-sided die. You roll once. If you keep that roll, you are paid the dollar amount of the face showing ($1 to $6). If you do not like your first roll, you can pay $0.50 to roll a second time, and you must keep the second roll. What is the fair price to play this game, and what is your optimal strategy?
Show the sample answer
Sample answer. Example answer written by our team. The applicant is invented. I would work backward from the second roll to find my decision threshold, then calculate the overall expected value. On a single roll of a fair six-sided die, the expected value is the average of 1 through 6, which is $3.50. If I choose to reroll, I have to pay $0.50, so the net value of taking the second roll is $3.50 minus $0.50, which equals $3.00. That gives me my stopping rule on the first roll. If I roll a 4, 5, or 6 on my first try, that payout is strictly greater than $3.00, so I should keep it. If I roll a 1 or a 2, that is less than $3.00, so I should pay the $0.50 and reroll. If I roll a 3, I am indifferent because keeping $3.00 equals my net expected value of $3.00 from rerolling, so I will keep the 3 to reduce variance. Now I can average the six equally likely first-roll outcomes under that strategy: for rolls 1 and 2, my expected net payout is $3.00 each. For rolls 3, 4, 5, and 6, I keep the face value. Summing those six outcomes gives 3 + 3 + 3 + 4 + 5 + 6, which is 24. Dividing 24 by 6 gives a fair game price of $4.00.
It states the backward-induction structure in the very first sentence, explicitly accounts for the $0.50 reroll cost before setting the threshold, notes the variance tie-breaker on rolling a 3, and walks through the numerator (24) and denominator (6) out loud so the interviewer can follow every step.
How to prepare for online assessments and technical interviews
How to prepare for online assessments and technical interviews
Use these five timed drill routines written by our team to prepare for screening puzzles and live technical conversations with traders, researchers, or engineers.
1. Exact fraction and
Exact fraction and integer arithmetic under a 3-minute timer
Solve probability and expected value puzzles on scratch paper without rounding to decimals until the final step. Keep numerators and denominators in simplified fractions (such as 14/21 = 2/3 and 6/8 = 3/4 in SIG's rain-and-pain table) so you keep exact precision across multi-step branches.
2. Two-by-two contingency ta
Two-by-two contingency table checks
Whenever a puzzle gives you joint counts or diagnostic test rates, draw a four-cell grid immediately and calculate both column conditional probabilities before stating whether two events are positively or negatively correlated.
3. Two-minute decision tree
Two-minute decision tree pruning
Set a two-minute timer on multi-stage dice, card, or coin games. Group identical states together and work backward from the final round instead of writing out every path.
4. Live market-making quote
Live market-making quote drills
Pick an uncertain quantity or probability event, quote a two-sided bid-ask spread out loud, and have a partner feed you one piece of new information. Move your midpoint immediately by the full Bayesian update so you overcome anchoring bias.
5. Role-specific coding and
Role-specific coding and modeling drills
For Operations Analyst roles, practice data manipulation in SQL, VBA, and Python. For Software Developer or Quantitative Research roles, practice debugging, system architecture, and quantitative modeling where traders in the same building are your end users.
Work through expected value and Bayes' theorem prompts out loud
Soreno's AI interviewer asks probability, expected value, Markov state, and decision-science questions by voice, listens to how you reason step by step, and scores your math and clarity. Soreno is not affiliated with Susquehanna International Group.
Questions people ask
Why does SIG focus on poker and strategy games in recruiting and training?
SIG uses poker to teach new traders how to make decisions under uncertainty, evaluate the expected value of a trade, and price risk, and it uses the trick-taking betting game Wizard to introduce job candidates to gaming at the firm.
What decision science concepts and cognitive biases does Susquehanna test?
SIG highlights four specific biases that hurt trading accuracy: confirmation bias (such as illusory correlation in a 29-day rain and pain table), anchoring and insufficient adjustment, availability bias (such as the gambler's fallacy after 10 heads in a row, a 1-in-1,024 event), and framing effects on risk tolerance.
Do I need prior finance experience to get a trading internship at SIG?
No. Candidates are hired without prior finance experience and learn options theory, market making, and decision science during the 10-week internship through daily classroom education, mock trading, and mentorship on the trading desk.
What does a Quantitative Trading Intern do during the 10-week summer at SIG?
Quantitative Trading interns split the 10 weeks into a five-week morning desk rotation and a five-week afternoon desk rotation, learning how traders make markets on options, stocks, and ETFs. The other half of each day is spent in options theory classes and real-time mock trading sessions.
How can freshmen and sophomores get an early introduction to SIG's interview process?
SIG runs Discovery Programs in trading and technology for students who are not yet eligible for its internship program. Discovery Programs introduce students to the firm's culture and provide an early introduction to the interview process for internships.