Jump Trading's Superday distills to four rounds in one day: two quant or math, one coding, one behavioral.
You are quoting a two-sided market in a futures contract, 100 lots on each side. Over five minutes you are lifted on your offer eight times in a row and never once hit on your bid, while the broader market has barely moved. What is most likely happening, and specifically what do you do to your quotes and your position?
Take a standard chessboard and count every square whose sides lie along the grid lines — not just the 64 unit cells, but squares of every size. How many are there? As a follow-up, count all the grid-aligned rectangles the board contains.
trucks enter a long single-lane mountain road, one behind another. Each truck has its own preferred cruising speed, and all speed orderings are equally likely. Overtaking is impossible, so a truck that catches up to slower traffic settles in behind it, and the trucks eventually travel as separated platoons. What is the expected number of platoons?
Casual, on background and motivation
Content depends on the track you chose
Two quant, one coding, one behavioral
Post-offer talks decide which desk you join
Practice a live-format mock interview
Five canvases numbered 1 through 5 sit stacked in random order, face down, and four easels stand in a row. You reveal canvases one at a time; each revealed canvas must immediately be placed on a still-empty easel of your choice and can never be moved again. The game ends the instant all four easels are occupied — the fifth canvas is never seen. You win if the easels read in strictly increasing order left to right. Playing optimally, what is your probability of winning?
and are independent standard normal random variables. What is ? Then what is ?
Cars pass a roadside sensor as a Poisson process with unknown rate per second, so the gaps between arrivals are independent exponentials. You start watching at time zero. The first car passes at seconds and the second at seconds. (a) What is the maximum-likelihood estimate of ? (b) Is that estimator unbiased? If not, produce one that is.
Two traders play for a fixed pot. Trader A names an integer out loud; then trader B, having heard it, names a different integer. Two fair dice are rolled and whoever's number is closer to the sum takes the whole pot. What number should A name, and what is A's chance of winning against a B who plays optimally?
You and a rival flip a fair coin repeatedly; the first to win 2 flips wins the match and the loser pays the stake, which starts at \$1. You alone hold a one-time right to double the stake to \$2, which you may exercise before any flip. When you exercise it, your rival either accepts (the match continues for \$2) or folds immediately and pays the current \$1. Your rival plays optimally. What is that right worth to you?
Six distinct numbers are arranged in a uniformly random order. You run exactly one left-to-right pass of bubble sort: compare positions 1 and 2 and swap if they are out of order, then positions 2 and 3, and so on through positions 5 and 6. What is the probability the list is fully sorted afterwards?
A strategy backtests at a Sharpe ratio of 3 over eight years, with a smooth equity curve. It has now been live for three months, is losing money, and has not had a single positive week. How would you investigate, and what single piece of evidence would best separate the competing explanations? Start by asking whether the live run is even consistent with the backtest.
Each second a token on a number line moves one step right with probability 0.3, stays put with probability 0.5, and moves one step left with probability 0.2, independently each second. After 100 seconds, what are the expected position and the standard deviation of the position? Approximately what is the probability the token is at or to the left of where it started?
Four back-to-back interviews in one day: two quantitative or math rounds of pen-and-paper probability and statistics, one coding round matched to your track, and one behavioral and fit round. Traders are pushed into futures microstructure and options theory, while quant researchers get applied statistics and questions on their own projects. It runs mainly at Jump’s Chicago headquarters.
A recruiter screen of about 30 minutes on background and motivation, then one or more technical phone interviews of 45 to 60 minutes. The technical content depends on your track: probability and expected-value brainteasers for traders, probability and statistics plus OLS and linear algebra for quant researchers, and data structures and algorithms in C++ for developers.
Roughly 3 to 5 weeks from the recruiter screen to a decision. Jump moves quickly once the onsite is finished, and an offer is often followed by team-match conversations to place you on a specific desk, since compensation is described as consistent across teams.
