Hudson River Trading and Tower Research Capital are nearly the same size and almost never the same job. HRT is one firm with one trading operation: about 1,490 people in March 2026 with 64% of them in North America (Revelio Labs), and a reported $6.4 billion of Q1 2026 trading revenue, up roughly 135% year on year (Hedgeweek, 12 May 2026, citing Bloomberg). Tower is a platform of independent trading teams, 1,539 people over the same month (Revelio Labs), a third of them in South Asia, where the desk you match into sets your work and much of your pay.
That structural difference decides almost everything else: how the interview is run, how the bonus is calculated, and how much downside you carry in year two. Here is what the public record actually supports.
The 30-second verdict
| HRT | Tower Research Capital | |
|---|---|---|
| Founded / HQ | 2002, New York; Jason Carroll and Prashant Lal (Wikipedia) | February 1998, New York; Mark Gorton and Alistair Brown (Wikipedia) |
| Structure | One firm, one trading operation, 14 offices (HRT careers, Sep 2026) | Partnership of independent trading teams incl. Latour and Limestone, 11 locations (Wikipedia) |
| Headcount, Mar 2026 | ~1,490 (Revelio Labs) | ~1,539 (Revelio Labs) |
| Where the people are | 64.0% North America, 14.5% Northern Europe (Revelio Labs) | 35.2% North America, 32.3% South Asia (Revelio Labs) |
| Job postings, 2026 | 107, down 76.7% year on year (Revelio Labs) | 86, up 24.8% year on year (Revelio Labs) |
| Reported results | Q1 2026: ~$6.4B revenue, ~$4.2B net profit, 70% EBITDA margin (Hedgeweek/Bloomberg, May 2026) | Does not publish firm-level revenue |
| Revenue per employee | $23.3M, vs Jane Street $18.3M and Virtu $3M (Rupak Ghose, May 2026) | Not disclosed |
| Grad hires per year | 80-120 from 30,000+ applications (Quantt estimate, May 2026) | 80-120 globally (Quantt estimate, May 2026) |
| First gate | OA, ~90 min, ~5-10% pass (Quantt estimate) | OA, 90-120 min, C++ performance for engineers (Tradermath, Jul 2026) |
| Year-one grad engineer | $300K-$500K (Quantt illustrative estimate) | $200K-$350K (Quantt illustrative estimate) |
| levels.fyi SWE median, 20 Sep 2026 | $506,872 (L1 $467K, L2 $507K, L3 $723K) | $209,000 across all locations |
| Glassdoor, retrieved Sep 2026 | 4.0 overall (148 reviews), WLB 3.1 | 4.1 overall (353 reviews), WLB 3.6 |
| Last 90 days | Hired Citadel's Jared Miller as head of debt investor relations (Hedgeweek, 15 Jul 2026) | Summer 2027 internships open across seven cities (Tower internships page, 20 Sep 2026) |
Quantt labels its pay bands illustrative estimates drawn from anecdotal reports, and the HRT financials are Bloomberg's reporting from unnamed sources, described as preliminary. Read the table as direction, not as offer letters.
Interviews: one loop versus an audition for two or three teams
The stages look almost identical, and they grade different things.
HRT sorts you into a track before any technical screen: quantitative researcher, algo developer, software engineer or quant trading. Tradermath's June 2026 guide describes a HackerRank online assessment of three problems at LeetCode medium to hard, one or two technical phone screens split between probability and live coding, then an onsite of about five one-on-one rounds covering coding, probability, strategy games and data analysis, over roughly four to eight weeks. Its example questions are a good picture of the taste: the classic Romeo-and-Juliet problem, where two people arrive uniformly at random in a one-hour window and each waits fifteen minutes; a 45-minute live poker simulation to identify the best hand; and reported rounds on order books, lock-free concurrency and arbitrage detection. Quantt's May 2026 HRT guide adds that offers usually arrive within two weeks of the onsite.
Tower's funnel runs on a similar timetable, but the onsite is doing a second job. Tradermath's July 2026 guide describes a 90 to 120 minute OA (two or three algorithmic problems in C++ with an explicit emphasis on performance for engineers; probability, statistics and data analysis for researchers and traders), one or two 60-minute phone screens, then a super day of five to six interviews in which you meet two or three different trading teams, followed by a team match over several days after the offer. Its listed engineering questions are systems questions rather than puzzle questions: a web-request throttler, TLB hits and misses, copy-on-write page counts, order-book design, the C++ memory model. On the quant side it lists asymmetric random-walk hitting times and Fermi estimation, including the surface area of a tennis ball.
The practical difference: at HRT you are convincing one firm. At Tower you are convincing several teams with different priorities, and the guide is explicit that candidates who can speak to multiple teams do better than candidates who pre-commit to one. Tradermath also notes that at both firms almost every question has follow-up variants, so the graded object is your reasoning, not the closing answer. If you want to drill that under time pressure, Applr's HRT quantitative researcher mock runs the probability-then-implementation pattern these loops use: /en/ai-mock-interview/hrt/quantitative-researcher.
Pay: one bonus pool versus your desk's P&L
The headline estimates favour HRT. Quantt's May 2026 guides put first-year graduate engineers at $300,000 to $500,000 at HRT and $200,000 to $350,000 at Tower. levels.fyi on 20 September 2026 shows an HRT software engineer median of $506,872, with L1 at $467K, L2 at $507K and L3 at $723K.
Tower's levels.fyi software engineer median is $209,000, and that number is close to useless for a US or London candidate. Revelio Labs put 32.3% of Tower's headcount in South Asia in March 2026 against 35.2% in North America, and Tower's own office list runs through Gurgaon and GIFT City. levels.fyi says so itself, quoting a range from $53,765 for a software engineering manager in India to $562,500 for a US data scientist in the same dataset. For a like-for-like read, Glassdoor's London figures, 43 salaries as of February 2026, run from $165,568 for an infrastructure developer to $374,193 for a quantitative trader.
The structural point matters more than either median. Tower's careers page describes independent trading teams operating across asset classes, and Tradermath's guide says compensation varies meaningfully by desk. A Tower employee answering a March 2023 Blind thread put numbers on the tail risk: a senior engineer had been offered a quant developer seat on a brand-new team with a base matching big tech and no guaranteed bonus even in year one, with team P&L access expected only after eighteen months to two years. The Tower respondent described that as roughly the risk profile of joining a three-person startup, binary within two years, with the bonus dictated by the portfolio manager because a new pod has no P&L buffer to pay from, and argued you should price that risk into the offer rather than hope.
HRT's economics point the other way. The Bloomberg-reported Q1 2026 figures are firm-level, not desk-level: about $6.4 billion of trading revenue, roughly $4.2 billion of net profit, about $4.5 billion of EBITDA on a 70% margin, and roughly $20 billion of net trading capital deployed (Hedgeweek, 12 May 2026). A May 2026 analysis by Rupak Ghose put HRT's revenue per employee at $23.3 million, ahead of Jane Street's $18.3 million and Virtu's $3 million, and its US cash equities share second only to Citadel Securities at around 20%. Full-year 2025 trading revenue was reported at $12.3 billion, and the firm is described as spending around $1 billion a year on artificial intelligence (Wikipedia, retrieved September 2026). You are joining a single, very profitable book rather than betting on one desk inside it.
Culture, hours and the hiring signal nobody reads
Glassdoor's side-by-side comparison, retrieved September 2026, is one of the rare places the two firms diverge in Tower's favour: Tower 4.1 overall from 353 reviews against HRT 4.0 from 148, work-life balance 3.6 against 3.1, culture and values 4.0 against 3.8. HRT leads on senior management, 3.6 against 3.3. A Tower employee on Blind in August 2024 gave the caveat that makes those averages readable: work-life balance at Tower depends heavily on the team, and there is a real difference between a "tower core" quant developer and a quant developer embedded in a trading team. A firm-wide average of a federation is an average of very different jobs.
The most useful number for anyone applying this autumn is the hiring one. Revelio Labs counted 107 active HRT postings in 2026, down 76.7% year on year, while Tower's rose 24.8% to 86. HRT is posting far less in the middle of a record year, which is what operating leverage looks like from the outside and what a narrower funnel looks like from the inside. HRT's recent moves are financial rather than headcount: it completed a $2.9 billion refinancing of a $2.35 billion loan maturing in 2030, returned for another $600 million, and in July 2026 hired Citadel's Jared Miller into a newly created head of debt investor relations role (Hedgeweek, 15 July 2026).
Tower's history also carries a compliance line worth knowing before an interview: in 2019 the firm paid $67.4 million to settle CFTC spoofing charges over E-mini futures, and its Latour subsidiary was fined $16 million in 2014 for net capital rule violations (Wikipedia, retrieved September 2026). It is old news internally, but it is public, and "why Tower" answers that ignore how the firm is actually organised land badly.
Which should you choose?
- You want the highest expected new-grad package and a single clear employer: HRT. Every public estimate is higher, the loop is one firm's decision, and the reported firm-level economics are extraordinary.
- You want to own a strategy early and can price risk: Tower. Joining a team is closer to joining a small business with shared infrastructure, with the upside and the two-year downside the Blind insider described.
- You are a systems engineer more than a puzzle solver: Tower's OA and onsite lean into C++ performance, memory models and order-book design. HRT tests the loop between maths and code.
- You need sponsorship: check the office, not the logo. Most HRT seats are US seats in the H-1B lottery, with London and Dublin the main alternatives; Tower's 2027 internship page on 20 September 2026 listed New York, Chicago, London, Singapore, Hong Kong, Shanghai and Sydney. Tower's South Asian headcount is real employment but not a route to a US or UK visa.
- You value predictable hours: Glassdoor's work-life balance gap (3.6 to 3.1, retrieved September 2026) favours Tower, with the caveat that it varies by team.
Prepare for both
Tower's Summer 2027 internships are open now, so the sequencing this month is straightforward: apply to Tower while the cohort is live, and treat HRT's narrower posting count as a reason to be early rather than a reason to skip it.
- Practise the HRT loop's probability-into-code pattern: HRT quantitative researcher mock
- Track which graduate applications are open and closing: New Grad 2027 tracker
- Check your resume parses before a Greenhouse ATS sees it: free resume checker
- If HRT is your target, the two nearest comparisons are already written up: Jump Trading vs HRT and Jane Street vs HRT
Sources
- Hedgeweek, "Hudson River Trading reports record Q1 revenue", 12 May 2026
- Hedgeweek, "Hudson River Trading hires Citadel exec for new debt investor relations role", 15 July 2026
- Rupak Ghose, "Hudson River Trading Q1 revenues up 135% and profits up 175%", 11 May 2026
- Revelio Labs, Hudson River Trading headcount, March 2026
- Revelio Labs, Tower Research Capital headcount, March 2026
- Wikipedia, Tower Research Capital, retrieved September 2026
- Wikipedia, Hudson River Trading, retrieved September 2026
- Quantt, Hudson River Trading interview process, 3 May 2026
- Quantt, Tower Research Capital interview process, 12 May 2026
- Tradermath, Hudson River Trading interview guide, 8 June 2026
- Tradermath, Tower Research Capital interview guide, 30 July 2026
- levels.fyi, Hudson River Trading salaries, retrieved 20 September 2026
- levels.fyi, Tower Research Capital salaries, retrieved 20 September 2026
- Glassdoor, Tower Research Capital vs Hudson River Trading comparison, retrieved September 2026
- Glassdoor, Tower Research Capital salaries in London, February 2026
- Blind, "tower research trading team offer", 7 March 2023
- Blind, "tower research capital", 27 August 2024
- Tower Research Capital, internships, retrieved 20 September 2026
- Hudson River Trading, careers, retrieved 20 September 2026
Frequently asked questions
- Is HRT harder to get into than Tower Research?
- On the public estimates, yes at the first gate. Quantt's May 2026 guide says HRT receives over 30,000 applications for roughly 80 to 120 graduate seats a year and puts online-assessment pass rates at around 5 to 10%. Quantt's Tower guide estimates a similar 80 to 120 graduate hires but does not report an application count. The bigger difference is later: at Tower you also have to win a team, because the onsite spans two or three trading teams and a team match follows the offer.
- Which pays more for a new grad, HRT or Tower Research?
- HRT, on every public estimate, but none of them is official. Quantt's May 2026 guides put first-year graduate engineers at $300,000 to $500,000 at HRT and $200,000 to $350,000 at Tower, both labelled illustrative. levels.fyi on 20 September 2026 shows an HRT software engineer median of $506,872, with L1 at $467K and L2 at $507K. Tower's software engineer median on levels.fyi is $209,000, but that figure mixes New York with Gurgaon and GIFT City, so it understates a US or London offer.
- Why is Tower Research's levels.fyi median so low?
- Geography. Revelio Labs put 32.3% of Tower's roughly 1,539 employees in South Asia as of March 2026, against 35.2% in North America, and Tower's own office list includes Gurgaon and GIFT City. levels.fyi mixes those locations into one median and explicitly quotes a range from $53,765 for a software engineering manager in India to $562,500 for a US data scientist. Compare like with like: Glassdoor's London data, 43 salaries as of February 2026, ranges from $165,568 for an infrastructure developer to $374,193 for a quantitative trader.
- What is Tower's team match and does it affect pay?
- Tower runs a partnership of independent trading teams, and its careers page describes them that way. Tradermath's July 2026 guide says the onsite spans two or three teams, that a team match follows the offer, and that compensation varies meaningfully by desk. A Tower employee answering a March 2023 Blind thread described joining a brand-new pod as carrying the risk profile of a three-person startup, with no guaranteed bonus and a binary outcome within two years, because a new team has no P&L buffer to pay from.
- Is HRT still hiring after its record quarter?
- It is hiring, but posting far less. Revelio Labs counted 107 active HRT job postings in 2026, a 76.7% fall from the prior year, while Tower's postings rose 24.8% to 86 over the same period. That is despite HRT's reported Q1 2026 results: about $6.4 billion of trading revenue, roughly $4.2 billion of net profit and a 70% EBITDA margin (Hedgeweek, 12 May 2026, citing Bloomberg and preliminary figures). Record revenue with fewer open roles means a tighter funnel, not a looser one.
- Which is better for an international student needing sponsorship?
- Check the office, not the firm. HRT's careers page lists 14 offices and Revelio Labs put 64% of its people in North America, so most HRT roles are US roles competing in the H-1B lottery, with London and Dublin as the main alternatives. Tower lists 11 locations, and its 2027 internship page on 20 September 2026 showed New York, Chicago, London, Singapore, Hong Kong, Shanghai and Sydney. Tower's large Gurgaon and GIFT City presence does not help a US or UK visa case; it just makes the aggregate salary data look lower.