RepostedJP Morgan Chase logo

JP Morgan Chase · OA → Phone screen → Onsite

In process

JPMorgan Quant Intern Superday: 19 Experiences (Part 1 of 3)

Quant (general) · InternshipReported Nov 27, 2025
Shared byjanebeingjane

JP Morgan Chase quant research intern reports, 2023 to 2025: Superdays of three or four rounds on probability, regression and ML, plus short Python tasks.

At a glance

  • Process: HireVue or an online test, sometimes a live coding or first interview, then a Superday of three or four interviews, 45 minutes each where timed (02, 06, 08). Indian campus routes ran a math and coding test, then four technical rounds and HR (03, 04, 05).
  • Asked recently: probability, statistics or math in 8 of 8 current reports; some coding in 8; ML or regression in 5; resume and project talk in 6.
  • Difficulty: candidates described Green Book-style probability and short Python tasks rather than hard algorithms (02, 06, 07); one PhD candidate blamed fatigue, not tricky questions, for a rejection (06).
  • Practise first: Fibonacci Number, Integrate a Sampled Function, Estimate Pi from Sampled Points and Train a Regularized Boosted-Tree Regressor.

Coverage: JP Morgan Chase quant research, Quant Analytics and Markets QR internships, undergraduate to PhD, from online tests to Superday. Eight current reports from November 2023 to November 2025 (three campus reports from 2025, published June 2026), plus one August 2023 report under History. New York, the UK and Indian campus processes where stated.

In this series

  • Part 1 (current): reports 01–09, 2023–2026: regression and ML theory, short Python tasks and campus rounds.
  • Part 2: 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.

The full series holds 19 reports. The eight current ones are all here; the other eleven are older than November 2023 and sit under History in each part.

What came up

Every current report mixed probability or math with some code. Regression ran from OLS assumptions to solving it faster with a QR decomposition.

Topic Reports Last seen
Probability, statistics or math 8 of 8 Nov 2025
Coding (DSA, Python or a short task) 8 of 8 Nov 2025 (report date)
Resume, project or behavioral talk 6 of 8 Nov 2024 (report date)
ML methods or regression 5 of 8 Nov 2025
Puzzles 3 of 8 2025
Linear algebra 3 of 8 Nov 2024 (report date)
Python data structures or OOP 3 of 8 Nov 2024 (report date)
Stochastic processes 2 of 8 Nov 2025

A candidate can appear in more than one row.

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Fibonacci with dynamic programming 1 Nov 2024 (report date)
Integrate a numerical function in Python 1 Nov 2024 (report date)
Estimate pi with Monte Carlo 1 Oct 2024 (report date)
Merge sort time complexity 1 Oct 2024 (report date)

ML and regression

Question Reports Last seen
How would you train an XGBoost model? 1 Nov 2025
OLS assumptions, how to check them, and a proof of OLS 1 Nov 2025
Lasso versus Ridge; is the Lasso solution unique? 1 Nov 2025
What happens to the variance of beta when predictors are added? 1 Nov 2025
Given a dataset, which method predicts the target? 1 Nov 2025
Run a linear regression and calculate its coefficients 1 Nov 2024 (report date)
Matrix-form regression, sped up with a QR decomposition 1 Nov 2023

Probability, statistics and math

Question Reports Last seen
Derive the mean and variance of distributions 1 2025
Define a positive semidefinite matrix and prove it 1 Nov 2024 (report date)
Explain the Poisson distribution 1 Oct 2024 (report date)

Behavioral (BQ)

Question Reports Last seen
How would you react to negative feedback at work? 1 Nov 2024 (report date)
You flag a model problem and a trader ignores you: what next? 1 Nov 2024 (report date)

8 detailed candidate timelines, loops, and screens

01: Math and a short coding task over Zoom

Report date: November 27, 2025 · Role: Quant intern, undergraduate · Result: Rejected

  1. Two interviewers asked math technical questions and a short coding task over Zoom. Both reacted well.
  2. The rejection came about two weeks later. No prompts were shared.

02: A New York Superday on regression, Lasso and XGBoost

Interview date: November 20, 2025 · Report date: November 26, 2025 · Role: Quant intern, PhD candidate · Location: New York, US · Result: Pending

  1. A live coding invite came about a month after HireVue, and the Superday invite three days after that.
  2. The Superday had no live coding. It covered basic stochastic processes, Green Book probability questions, and statistics and ML.
  3. Questions they listed: OLS assumptions, how to check them and a proof of OLS; Lasso versus Ridge and whether the Lasso solution is unique; how to train an XGBoost model; which method they would use to predict a target from a given dataset; and what happens to the variance of beta when predictors are added.
  4. They felt confident, but still had no decision in December.

03: WSC campus: ML, NLP and a 20-minute number-theory problem

Interview year: 2025 (from the June 2026 publication) · Role: Quantitative research intern, WSC · Location: IIT Kharagpur campus, India · Result: Offer

  1. Test: a math section in three parts (probability, calculus and school math, and a part they tentatively called MAT) and a section with two medium-to-hard DSA tasks.
  2. Round 1: ML, NLP and probability. Round 2: harder probability, arrays and bit manipulation, and puzzles. Round 3: ML and NLP applications and the CV. Round 4: a number-theory problem that took about 20 minutes, statistics and probability, and a project.
  3. HR followed, then the offer. The summary counted four interviews; the narrative counts five with HR.

04: CIB campus: four technical rounds and HR

Interview year: 2025 (from the June 2026 publication) · Role: Quantitative research intern, CIB · Location: IIT Kharagpur campus, India · Result: Internship received

  1. An online test came first.
  2. Round 1: the CV and profile, including earlier data science research and software internships, plus basic math and DSA.
  3. Round 2: puzzles, probability and ML. Round 3: math and probability and statistics, including deriving the mean and variance of distributions on paper (which distributions was not said). Round 4: puzzles, probability and ML. Round 5: HR.
  4. Most problems were not named. The summary said four interviews; the narrative has four technical rounds plus HR.

05: MRGR campus: DSA first, then probability and linear algebra

Interview year: 2025 (from the June 2026 publication) · Role: Quantitative research intern, MRGR · Location: IIT Kharagpur campus, India

  1. CV screening, then a tightly timed math and coding test.
  2. Four interviews. The first was mostly DSA. Later ones mixed probability and statistics, DSA, puzzles, the CV and some linear algebra.
  3. Projects got only an informal overview. No coding task, probability setup or puzzle was shared.

06: Four UK off-cycle rounds, ending with Fibonacci

Report date: November 14, 2024 · Role: Quantitative Research off-cycle intern, STEM PhD candidate · Location: UK · Result: Rejected

The online Superday was four back-to-back 45-minute interviews. The candidate was moving into quant from another STEM field and had about a week and a half to prepare.

  1. How would they react to negative feedback at work? Then: how would they run a linear regression and calculate its coefficients?
  2. A long talk about their background, then more regression, Python data structures, time complexity and a Green Book question on a sum of random variables (the variables were not recalled).
  3. Define a positive semidefinite matrix and prove it; they could not finish the proof. Then how to integrate a numerical function in Python, practice: Integrate a Sampled Function. An expected maximum or minimum question followed, without its setup.
  4. A scenario: they find a problem in a model, tell a trader, and get ignored. Python structures again, then Fibonacci Number with dynamic programming and its time complexity.

They had no stochastic-model or options-pricing background, and those topics were not pursued. They put the rejection down to short preparation and fatigue rather than tricky questions.


07: Three Quant Analytics interviewers, three resume threads

Report date: October 27, 2024 · Role: Quant Analytics intern, PhD candidate · Result: Waiting

A Friday-morning Superday with three interviewers.

  1. An interviewer based in Jersey City asked about the graph neural network work on their resume, then basic data structures.
  2. In a shared plain text editor they wrote Python with lists and tuples. They made an OOP mistake, then a small slip in an easy, unnamed algorithm, which they fixed. They were also asked for merge sort's time complexity. Afterwards they worried that sharing only the editor window, not the whole desktop, was a problem.
  3. A senior interviewer asked how to estimate pi with Monte Carlo, practice: Estimate Pi from Sampled Points. An options-pricing follow-up was dropped once they explained their engineering background. They then explained the Poisson distribution and did not recognize another term they could not later recall.

08: Markets QR: matrix regression and a QR decomposition

Interview date: November 10, 2023 · Report date: November 16, 2023 · Role: Markets Quantitative Research intern, master's candidate · Result: Not reported

  1. OA in late September, a first interview in late October, a Superday invite in early November, and three Superday interviews on November 10.
  2. Topics: Python OOP, unnamed LeetCode-style algorithms, Green Book probability and stochastic processes, and detailed resume questions.
  3. The one concrete follow-up: write linear regression in matrix form, and explain how a QR decomposition improves the time to solve it.
  4. Two rounds went well and one badly, in their view. They asked whether an offer was still possible.

History (before November 2023)

H09: Two CIB rounds, two QR rounds and two offers

Report date: August 21, 2023 · Role: Quant research intern; also interviewed for CIB · Location: India · Result: Offers from both; chose QR

CIB interview 1

  1. Write a sorting algorithm on paper; they wrote quicksort.
  2. R-squared versus adjusted R-squared; define a p-value and the null hypothesis.
  3. Given a book's total errors, find the probability of a stated number of errors on one page. They named the Poisson distribution and the interviewer accepted it.
  4. Their credit-fraud project: parameters and problems, the large imbalance between real and fraudulent transactions, and why they chose SMOTE oversampling. They had removed redundant columns using VIF, so they were asked to write and explain VIF and multicollinearity.
  5. Bonds, calls, puts and an options case (details not given), and how the Silicon Valley Bank collapse compares with the 2008 crisis.

CIB interview 2

  1. Which project ML method had the best accuracy; explain SVM; explain the research project; explain DBSCAN and why not KNN. A sorting algorithm was asked again.

Quant research interview 1

  1. Introduce yourself and name a favorite data structure; explain linear regression; write the regression error function and interpret the feature weights; explain logistic regression and how its weights differ from linear regression's (with hints); why quant?

Quant research interview 2

  1. Introduce yourself; SVM and soft margins; the closed-form solution for linear regression; gradient descent and why use it there; logistic regression and its gradient; is there a closed form for logistic regression, and is gradient descent needed; explain DBSCAN; why quant again.

HR

  1. CIB or QR preference, and other offers. Asked about SQL in QR, they declined SQL questions; no SQL task was given.

Patterns across the reports

  1. Campus rounds blend math and code. The three Indian campus reports all paired probability and statistics with DSA (03, 04, 05); WSC added ML and NLP, MRGR added linear algebra.
  2. Superdays run three or four interviews. Three interviews in 07 and 08, four in 06. The campus processes had four technical rounds before HR (03, 04, 05).
  3. Green Book probability keeps coming up. Candidates named it in 02, 06 and 08.

Common questions

Does the JP Morgan Chase quant intern Superday have coding?

Usually a little. Report 06 had Fibonacci and numerical integration in Python, 07 wrote Python in a plain editor, and 08 had unnamed algorithms. The 2025 New York Superday had none (02).

What regression questions does JP Morgan Chase ask quant interns?

OLS assumptions and proof, Lasso versus Ridge and beta variance (02), computing coefficients (06), and matrix-form regression with QR decomposition (08).

How many rounds is the JP Morgan Chase quant research Superday?

Three in 07 and 08, four 45-minute rounds in 06. Campus processes had four technical rounds plus HR (03, 04).

How long does JP Morgan Chase take to reply after a quant Superday?

Report 01 heard about two weeks later (a rejection). Report 02 was still waiting in December after a November 20 Superday.

Keep practicing

In this series

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

No comments yet — be the first to share what you know.