JP Morgan Chase · OA → Phone screen → Onsite
In processJPMorgan Quant Modeling Superdays: 7 Experiences (Part 2 of 3)
JP Morgan Chase quant intern reports from 2021 and 2022: three 45-minute Superday rounds on probability, modeling cases like card churn, and light coding.
At a glance
All seven reports here are from 2021 and 2022, more than three years old, so the whole part sits under History. For current reports, see Part 1.
- Process: HireVue with three recorded coding tasks, sometimes a 45-minute phone or Zoom screen, then a Superday of three 45-minute interviews (H13, H14, H15). An Indian campus route ran a sectioned test, three technical rounds and HR (H16).
- Asked in this period: probability, statistics or math in 6 of 7 reports; coding or data structures in 6; modeling or data analysis in 5. Named tasks include a credit-card churn case (H11), a sine-power integral (H11), a handwritten GCD and choosing K in kNN (H14).
- Difficulty: one candidate ran out of time on an interest-rate swap (H11); another stalled on a probability inequality and a recalled O(1) restriction and was rejected (H14).
- Practise first: Predict Card Customer Departure After the First Year, Greatest Common Divisor and Select the Neighborhood Size for kNN.
Coverage: JP Morgan Chase Quant Analytics and Quant Research internships, including summer-associate applicants from undergraduate to PhD. Six Superday reports and one Indian campus process, July 2021 to February 2022. New York and India where stated.
In this series
- Part 1: reports 01–09, 2023–2026: regression and ML theory, short Python tasks and campus rounds.
- Part 2 (current): reports 10–16, 2021–2022: a credit-card churn case, kNN, a handwritten GCD and probability.
- Part 3: reports 17–19, 2018–2021: stationarity, Black–Scholes and light coding.
What came up
Most of these Superdays split the day between a math interviewer and people who cared about data and modeling. Counts cover the 7 reports in this part.
| Topic | Reports | Last seen |
|---|---|---|
| Probability, statistics or math | 6 of 7 | Feb 2022 |
| Coding or data structures | 6 of 7 | Feb 2022 |
| Resume, behavioral or team talk | 5 of 7 | Jan 2022 (report date) |
| Modeling, regression or data analysis | 5 of 7 | Jan 2022 (report date) |
| Finance or derivatives | 2 of 7 | Feb 2022 |
| System design | 1 of 7 | Feb 2022 |
A candidate can appear in more than one row.
Questions we found
Algorithms / Data Structures
| Question | Reports | Last seen |
|---|---|---|
| Find the largest and second-largest values | 1 | Oct 2021 (report date) |
| Greatest Common Divisor, on paper | 1 | Oct 2021 (report date) |
| Explain a stack and its time and space costs | 1 | Oct 2021 (report date) |
| Explain a sorting algorithm and its time and space costs | 1 | Oct 2021 (report date) |
| SQL versus MongoDB; merge sort and quicksort complexity | 1 | Sep 2021 (report date) |
| Polymorphism and inheritance, with examples | 1 | Jul 2021 (report date) |
ML cases and modeling
| Question | Reports | Last seen |
|---|---|---|
| Predict whether a credit card customer leaves after a year | 1 | Jan 2022 (report date) |
| Plan a two-sample experiment and choose a test | 1 | Oct 2021 (report date) |
| How kNN works and how to choose K | 1 | Oct 2021 (report date) |
Probability, statistics and math
| Question | Reports | Last seen |
|---|---|---|
| Recurrence for the integral of sin(x)^n from 0 to pi | 1 | Jan 2022 (report date) |
| For X, Y each U(0,1), find P(XY < 0.5) | 1 | Jan 2022 (report date) |
| Name Monte Carlo variance-reduction techniques | 1 | Jan 2022 (report date) |
| Expected die rolls to see every face | 1 | Jan 2022 (report date) |
| Does x_n = sqrt(x_(n-1) + 2) converge, and to what? | 1 | Jan 2022 (report date) |
| Interpret a confidence interval and a p-value | 1 | Oct 2021 (report date) |
Finance
| Question | Reports | Last seen |
|---|---|---|
| Outline an interest-rate swap calculation | 1 | Jan 2022 (report date) |
Behavioral (BQ)
| Question | Reports | Last seen |
|---|---|---|
| Why JP Morgan, and which recent company news did you follow? | 1 | Oct 2021 (report date) |
| Explain statistical results to a nontechnical person, with an example | 1 | Oct 2021 (report date) |
History (before November 2023): 7 detailed candidate timelines
H10: FX e-trading questions, then a relaxed Treasury talk
Interview date: February 18, 2022 · Report date: February 24, 2022 · Role: Quant Analytics intern, master's candidate · Result: Not reported
- The first Superday interview was with a New York FX e-trading team: more than ten technical questions on data structures, math, finance and system design.
- With no system design background, the candidate struggled there. No design prompt was shared.
- A Treasury interviewer then held a relaxed conversation of about 40 minutes.
H11: Credit-card churn, a sine integral and a nested square root
Report date: January 25, 2022 · Role: Quant intern, PhD candidate · Result: Passed (offer acceptance not stated)
The three-round Superday started at 8:30 a.m. Eastern, which was 5:30 a.m. for this West Coast candidate, with no choice of slot.
Round 1: modeling and finance
- What data and model would predict whether a customer leaves after a year with a JP Morgan Chase credit card?
- How to calculate an interest-rate swap. Time ran out, so they outlined the approach only.
Round 2: probability and math
- For X and Y each U(0,1), find P(XY < 0.5). Whether they were independent was not stated.
- Integrate sin(x)^n from 0 to pi and derive a recurrence.
- Name variance-reduction techniques for Monte Carlo.
- The expected number of rolls of a six-sided die to see every face.
Round 3: a limit and the MRGR team
- Does x_n = sqrt(x_(n-1) + 2) have a limit, what is it, and why? The starting value was not given.
- The rest was behavioral talk and a description of the MRGR team's work. The second and third interviewers first asked what had already been covered, which the candidate read as a sign of a shared question list.
H12: One math interviewer and two data-analysis interviewers
Report date: November 19, 2021 · Role: Summer Quant Analyst 2022, PhD candidate · Result: Not reported
- A three-interviewer Superday on a Tuesday. One person asked math; the other two focused on data analysis.
- In a follow-up the candidate described a case-style discussion and basic data-analysis algorithms, with no hands-on computer work. The case and algorithms were not shared.
- They later got a request for more personal information and asked what it meant.
H13: Three recorded coding tasks and three 45-minute rounds
Interview month: October 2021 · Role: Quantitative analyst summer associate intern · Location: New York, US · Result: Offer declined
- HireVue had three coding tasks with recorded explanations.
- A 45-minute phone or Zoom interview, then a Superday of three 45-minute one-on-one interviews. The process took three to four months.
- No coding problem or later technical question was named.
H14: A two-sample study, a handwritten GCD and choosing K
Report date: October 19, 2021 · Role: Quantitative Research Summer Analyst 2022, undergraduate statistics student · Result: Rejected
- Online coding: three tasks with recorded explanations: string or array manipulation, a DP task they thought came from LeetCode, and a long third task they did not finish. None was described well enough to identify.
- The Superday invite came about two months after applying in early May, with a week to prepare. Three 45-minute rounds.
Round 1
- A research project: which model, what it predicted and why it was useful. The interviewer cut in before they finished.
- Interpret confidence intervals and p-values, then plan an experiment for two samples, pick a test and explain why.
- Explain a stack and its time and space costs. Then find the largest and second-largest values in an unsorted array (K = 2) and say what to track. A later restriction was unclear: they recalled O(1) time and space, could not meet it and got no hint. O(1) time is not possible when every element must be read.
- Basic SQL concepts, with no SQL written.
Round 2
- Why JP Morgan? After a culture answer, which recent company news did they follow and why. They struggled with the follow-up.
- How would they explain statistical results to a nontechnical person, and when had they done it? The interviewer did not seem satisfied with "plain English and practice".
- A card probability problem (rules not given).
- Greatest Common Divisor, written on paper. They held the notebook to the camera, read the code aloud, and repeated cases when the interviewer could not hear.
Round 3
- A probability problem was dropped when they could not solve an inequality. The interviewer went back to a project and asked how kNN works and how to choose K, then hash maps and a sorting algorithm with its time and space costs.
- HR first pointed to a Monday decision; the rejection came about three weeks later.
H15: Statistics, programming and behavioral, one team each
Report date: September 23, 2021 · Role: Quant intern, summer 2022 · Result: Offer
- Applied online without networking. A Pymetrics result from the year before was reused.
- Three coding questions: 30 minutes each, then three minutes to explain the code, with two recording tries. They used R. Behavioral questions followed, ending with a recent market event.
- About a month later, a Superday of three one-on-one 45-minute interviews with five-minute breaks. Different teams covered statistics, programming and behavioral.
- Resume questions included SQL versus MongoDB (both were in a course project), merge sort and quicksort complexity, modeling details, data preparation and missing values. They felt the job sounded more like data analysis than the quant title.
- The offer arrived a little under two weeks later.
H16: Transforms, Markov chains and OOP on campus
Report date: July 10, 2021 · Role: Quantitative research intern · Location: IIT Roorkee campus, India · Result: Offer
- An ECE student. The test had separately timed math, aptitude and programming sections, about one or two minutes a question.
- Four live interviews: three technical and HR, moving from algebra to probability, programming and puzzles.
- Technical topics: Fourier and Laplace transforms, the Newton–Leibniz theorem, regression, Markov chains and probability, and polymorphism and inheritance with examples. The exact problems and puzzles were not shared.
Patterns across the reports
- Three interviews was the norm. H11 to H15 each had three interviewers or rounds; H10 described two team conversations.
- Projects steered the technical questions. H14 went back to a project for kNN, and H15 drew database and modeling questions from a course project.
- Recorded coding came first. H13, H14 and H15 all had three recorded coding tasks before the Superday.
Common questions
What was the JP Morgan Chase quant intern Superday like in 2021?
Three 45-minute interviews, often split between math, data or modeling and behavioral (H12 to H15). Report H11 started at 8:30 a.m. Eastern with no choice of slot.
Did JP Morgan Chase quant interns code during the Superday?
Some did: a GCD on paper and an array task in H14. H12 had no hands-on computer work, and H15 discussed complexity rather than writing code.
What was the JP Morgan Chase quant HireVue like?
Three coding tasks with recorded explanations (H13, H14, H15). In H15 each task had 30 minutes plus three minutes to explain.
Keep practicing
- Predict Card Customer Departure After the First Year and Integral of a Sine Power: from H11.
- Greatest Common Divisor, K Largest Integers in Descending Order, Design a Two-Sample Mean Comparison and Select the Neighborhood Size for kNN: from H14.
- JP Morgan Chase problem bank: every JP Morgan Chase question on FastPrep (sign-in needed).
- JP Morgan Chase interview page: rounds, question types and recent questions.
- Data and AI intern reports and tech intern reports: the other JP Morgan Chase intern tracks.
- Experience feed: the latest interview write-ups.
In this series
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.