01 · Quant Career Path
Written for people with STEM/finance backgrounds who want to enter the quant industry. This article covers how to land a job at a quant institution, not how to trade with a systematic strategy — for the latter, see Quant Practice.
The quant industry has been one of the fastest-expanding tracks in finance over the past decade: entry-level salaries at top funds have long ranked in the market's first tier (actual market conditions prevail), and what comes with it is an extremely high bar and an extremely low acceptance rate. This article explains roles, skills, resumes, interviews, channels, and transition realities in one pass — no motivational fluff.
Roles in the Quant Industry
| Direction | Representative firms (public common knowledge) | Work | Salary structure basics |
|---|---|---|---|
| Buy-side · quant funds | Top names like Ubiquant (Jiukun), High-Flyer, Minghong, Yanfu; plus many small and mid-sized funds | Developing and iterating strategies (factors, machine learning, CTA, high-frequency), managing live capital; performance directly drives bonus | Base salaries rank near the top across industries (from hundreds of thousands to millions RMB); bonuses are a large share and highly volatile, with huge gaps between top and mid-tier firms |
| Buy-side · proprietary desks | Broker prop desks, futures prop desks, bank prop (some) | Trading with your own or the firm's money, no external client pressure, pursuing absolute returns | Fixed salary + performance commission; aggressive prop shops offer large bonus upside |
| Sell-side · financial engineering | Financial engineering teams at major broker research institutes | Writing research reports (factor stock selection, derivatives pricing, weekly quant notes), serving institutional clients like mutual funds and insurers — this is "sell-side service", not direct profit-making | Broker compensation system (base + bonus); top brokers pay better than average ones but generally below top buy-side funds |
| Foreign firms | Two Sigma, Citadel and other firms operating in China (subject to actual public information) | Equal weight on research and development; more internationalized processes and standards; some roles involve postings / full English | Tied to global systems; overall pay sits at the very top of the industry, though China-region dynamics have shifted considerably in recent years |
📖 Buy-side vs Sell-side
One-line distinction between buy-side vs sell-side: the buy-side makes money with its own capital (profits go to itself/its clients); the sell-side trades research for commissions (serves clients to make money). The buy-side is judged by performance; the sell-side by reputation and service quality.
Skill Stack Requirements
Whether buy-side or sell-side, core skills for quant roles fall into four blocks:
Mathematics
- Probability theory and mathematical statistics (random variables, distributions, hypothesis testing, regression) — the foundation under the foundation; about 60% of interview questions come from here.
- Linear algebra (matrix operations, eigenvalues, covariance) — the language of factor models and portfolio optimization.
- Calculus and introductory optimization — used for solving strategy parameters.
Programming
- Python is mandatory: pandas/numpy data processing, multiprocessing, unit testing, code standards. Research roles use Python almost exclusively.
- C++ is a plus: high-frequency/low-latency roles require C++, and it's essential for developer roles — for research roles it's "you can get in without it, but you're stronger with it".
- SQL, Git, and basic Linux operations are assumed.
Machine Learning
- From classic models (linear regression, decision trees/GBDT, clustering) to deep learning (applying CNN/LSTM/Transformer to factors, depending on the firm).
- The core is "solving financial problems with machine learning": feature engineering, overfitting prevention, out-of-sample validation — far more important than knowing how to tune hyperparameters.
- Market directions: fundamental factors, price-volume factors, CTA (trend/cross-section), statistical arbitrage.
Finance Knowledge
- Derivatives pricing basics (Black-Scholes formula, Greeks), trading rules (price limits, T+1, margin), market microstructure (order book, impact cost).
- The knowledge barrier for finance is relatively low compared to math and programming, and can be filled after joining — most institutions prioritize math and programming in assessments.
Skill Self-Assessment Table
| Skill | Beginner (can apply for internships) | Competent (can apply for campus recruiting) | Proficient (can apply for full-time research/dev) | My status |
|---|---|---|---|---|
| Probability/statistics | Can compute expectations, variance, common distributions | Random variable transformations, CLT, hypothesis testing | Stochastic processes (random walks/Markov/Brownian motion) | ☐ |
| Linear algebra | Can do matrix operations | Eigendecomposition, covariance matrices | Expressing portfolio optimization with matrices | ☐ |
| Python | Can write scripts, basic pandas | Can independently complete data cleaning and strategy computation | Engineering: classes, tests, performance optimization | ☐ |
| C++ (bonus) | Never written it | Can write simple programs | Understands memory model, can write low-latency components | ☐ |
| Machine learning | Knows names of common models | Can run classification/regression end-to-end with sklearn | Can design features, evaluate overfitting, do out-of-sample validation | ☐ |
| Finance knowledge | Knows what stocks/futures/options are | Can read candlestick charts, understands T+1/margin | Can derive Black-Scholes, understands market microstructure | ☐ |
| Projects | None | One complete backtest project | A reproducible research repo on GitHub (with report) | ☐ |
| Competitions | None | Participated in Kaggle/simulated competitions | Verifiable rankings or model-performance evidence | ☐ |
Self-assessment rule: check off every row before applying — far more effective than blindly mass-applying. The vast majority of campus-recruiting candidates fail on "can't solve probability problems + can't explain their projects clearly".
Where Campus Recruiting Fails
The vast majority fail on "can't solve probability problems + can't explain their projects clearly". These are the two most honest gates in quant job hunting; no resume polish gets you past follow-up questions.
Time Estimates for Skill Preparation (common-knowledge ranges)
| Starting point | To "internship-ready" | To "campus-recruiting-ready" | Notes |
|---|---|---|---|
| Math/statistics major | 1-3 months to pick up Python | 6-12 months (projects + ML) | Math is your home turf; fill in engineering |
| Computer science major | 2-4 months to pick up probability/statistics | 6-12 months (statistical intuition + projects) | Programming is your home turf; fill in math |
| Finance background | 3-6 months to pick up programming | 12-24 months (the gap is mostly math) | You have the finance knowledge; both math and programming need filling |
| Pure cross-discipline | 6-12 months | 18-36 months (long-termism required) | Longest timeline; prioritize the competition + project route |
💡 Preparation Time Depends on Effective Hours
Time estimates are reference only: what determines progress is not the calendar but effective study hours (roughly assuming 10-20 effective hours per week). Ignore any pitch promising "quant in three months".
Resume & Projects: What If You Have No Internship
No internship is the first hurdle for most cross-discipline applicants. Alternatives:
Personal Quant Project (Most Recommended)
- Build a complete backtesting framework or factor research repository, published on GitHub: README explaining your approach, data-fetching code, visualizations, and a conclusions report.
- What matters isn't flashy code but whether you can explain it clearly: what assumptions you made, how you prevented look-ahead bias, how you validated out-of-sample, how to interpret the results.
- Reference directions: dual moving average/Bollinger Band backtests (see Quant Practice), factor effectiveness research (IC/IR), pairs trading, CTA trend following.
Kaggle / Quant Competitions
- Kaggle finance competitions (e.g., the Optiver series): prove your "machine learning + financial data" ability; even a modest ranking has resume value.
- WorldQuant BRAIN: a free-to-register simulated quant platform where you can produce "alpha research reports" — one of the better-recognized "amateur portfolios" among Chinese quant funds.
- Simulated trading competitions hosted by brokers/exchanges/universities: a ranking is an excellent door-opener.
Open Source Contributions / Blogging
- Submit PRs and file issues on open-source quant libraries (backtrader, vn.py, klinecharts, etc.).
- Publish strategy research articles on WeChat/Zhihu/CSDN — interviewers actually read them; content quality matters more than follower count.
Resume Writing Points
- Put your tech stack somewhere visible; describe projects in three parts: "what you did → how you did it → what the result was".
- Don't write "expert"; write "used X to solve Y in scenario Z".
- Be careful with return-rate numbers: unverified return records can backfire; state the backtest period and assumptions.
Interview Process & Common Questions
| Stage | Content | Advice |
|---|---|---|
| Written test | Probability + coding questions (mostly LeetCode medium) + financial math at some firms | Grind problems; probability relies on long-term accumulation |
| First round (fundamentals) | Probability/statistics/coding basics; deep-dive into resume projects | You must be able to talk through each project's details for 30 minutes |
| Second round (strategy) | Design a strategy/factor for a given scenario; discuss overfitting and decay | Show your thinking process, not standard answers |
| Final round | Culture fit, stress tolerance, career plans, behavioral questions | Be sincere; don't fabricate experiences |
High-Frequency Question Types (Public Regulars)
- Coin-flip expectation: e.g., "expected number of flips until heads", "flips needed for two consecutive heads" — tests geometric distribution and Markov chains.
- Random walks: "does a random walk return to its origin?", "gambler's ruin" — tests stochastic-process intuition.
- Monte Carlo: describe/implement using random simulation to estimate expectations or price derivatives.
- Classic paradoxes: Monty Hall, birthday paradox — tests whether you've "seen the world".
- Behavioral: "why quant?", "how much drawdown can you tolerate?", "what do you think when a strategy stops working?"
Plenty of public interview write-ups and solutions exist online; systematically work through material at the level of "100 probability problems". The math depth tested in quant interviews isn't deep, but it must be fast and accurate.
The Truth About Quant Interviews
The math depth tested isn't deep, but it must be fast and accurate. This isn't selecting math geniuses; it's screening for people who can think while articulating and stay composed under pressure.
Three Classic Problems Walked Through (Public Questions, For Self-Testing)
| Problem | Approach | Answer |
|---|---|---|
| Fair coin: expected flips until first heads | Geometric distribution, expectation = 1/p | 2 |
| Expected flips until "heads-heads" appears consecutively | Set up an expectation equation (Markov state transitions) or recurrence, E = expected value | 6 |
| Symmetric ±1 random walk from 0: probability of ever hitting +1 | Martingale/unconditional symmetry analysis; symmetric random walks return to origin almost surely, and hit any integer | 1 |
- Question 1 tests whether you know the geometric distribution; question 2, whether you can set up a state equation; question 3, stochastic-process intuition — most interview questions reduce to these three types. Master the ideas before grinding.
- Prepare coding questions at LeetCode medium difficulty; focus on "writing it correctly, writing it fast, explaining the complexity clearly". No need to obsess over exotic hard problems.
High-Frequency Behavioral Questions
- "Why switch from your original industry to quant?", "why our firm?" — do your homework; be able to cite the firm's publicly known strategic direction and style.
- "Your most failed project/worst losing trade" — tests attribution ability: explain objective causes and improvements, without blame-shifting or self-deprecation.
- "How much drawdown can you take?", "what if a strategy fails?" — demonstrate "a stop-loss-and-iterate mechanism", not a "I can tough it out" attitude.
Job Hunting Channels
| Channel | Characteristics | Advice |
|---|---|---|
| Internship conversion | The main full-time pipeline at top institutions; formal-track acceptance rates are extremely low | Prioritize summer/long-term internships; on-the-job performance during internships counts more than credentials |
| Community referrals | Referrals via WeChat accounts, paid communities, GitHub circles convert far better than mass applications | Produce work first (projects/articles) so referrals carry weight |
| Headhunters | Suited for job-hopping with 1-2 years' experience; rare for fresh graduates | Keep in touch; don't trust verbal promises of "high pay" |
| Official channels | Company sites, Niuke, campus talks | Don't rely on a single channel; run several in parallel |
⚠️ The Reality of Campus Recruiting Quotas
Reality check: top quant funds hire only single-digit to low double-digit new graduates per year, and most headcount goes to internship conversions. For most people, "smaller fund first → build track record → move to a top firm" is more realistic than landing a top fund in one shot.
Transition Realities: How Intense Is the Competition
- Acceptance ratio: resume-to-offer ratios in campus recruiting at top quant firms are typically around one percent or lower (range of public discussion; actual market conditions prevail); internship-conversion slots are just as scarce.
- Who you're competing against: math, physics, and CS masters/PhDs from top Chinese universities (Tsinghua/Peking/Fudan/SJTU) and overseas schools form the main pool; most have "relevant degree + internship"; cross-discipline applicants need stronger portfolios to compensate.
- Beyond the halo: top-fund salaries come with high pressure — strategies get cut when they decay, performance reviews loom, and layoffs/restructuring uncertainty is constant; smaller firms pay ordinarily but with friendlier pace and faster learning.
- Cycle effects: the quant industry correlates strongly with market conditions and fundraising environment (hiring contracts during downturns and regulatory tightening); supply and demand swing sharply with cycles.
- Long-termism required: from "can write strategies" to "got hired" typically takes years of preparation; many people prepare two to three years for their first offer, and many fail and return to their original industry — that's normal. Decide your stop-loss line in advance.
Fresh Graduates / Experienced Hires / Interns: Different Plays Per Stage
| Stage | Your competitors | Play |
|---|---|---|
| In school (fresh grad) | Elite-degree peers with internships | Run competitions, projects, and internships in parallel; grab a smaller-firm internship first |
| Already employed (experienced hire) | Experienced candidates | Win on "verifiable projects + skill depth" rather than degrees; invest weekends, expect a longer runway |
| No degree but has work | Proof-first | WorldQuant BRAIN reports, GitHub repos, competition rankings are the main weapons |
| Internship conversion | Same cohort of interns | Solid delivery + proactive output during the internship matters more than credentials |
⚠️ If Employed, Don't Quit to Gamble
Realistic advice for the employed: keep your job and treat job hunting as your "second evening job" — the upside of quant job hunting is switching tracks if you succeed; the cost is months to years of spare time. Don't stake resignation on it.
Common Misconceptions
- Believing in "certificates": CFA/FRM carry far less weight in quant research hiring than "can solve problems, can code, has projects".
- Neglecting data skills: SQL, data cleaning, reconciliation mismatches — research roles spend much of their day on data engineering, and interviews often test it.
- Only reading interview write-ups without writing code: passing the written test but failing round two usually means "projects that can't survive follow-ups".
- Applying only to top firms: for every hire at a top firm, ten get hired at mid-tier ones; get in first, then advance.
Job Hunting FAQ
| Question | Answer |
|---|---|
| Can I get into quant without a top-school degree? | Yes, but let your work speak: at least two of — competition ranking, BRAIN reports, open-source projects, internships |
| Average math but strong coding — which track? | Developer roles (quant dev/systems engineer) demand less math than research roles; engineering ability is the ticket in |
| Is a master's degree mandatory? | Common-knowledge range: top-firm research roles mostly start at master's level, but developer roles and smaller firms are far more forgiving |
| Is quant a young person's game? | Both research and development careers are long-lasting; the "experience appreciates with age" shows up in strategy understanding and system architecture |
| Is not knowing C++ disqualifying? | Not for research roles; yes for developer roles. Different directions — no need to panic |
| Is it too late to start preparing now? | Quant is ability-driven; there's opportunity whenever you start. But if you've worked many years with no relevant accumulation, validate your commitment and cost-effectiveness part-time first |
Risk Warning
⚠️ Risk Warning
Quant job hunting is a textbook high-barrier, high-competition track with heavy survivorship bias. The high-salary stories circulating online concentrate on top roles at top firms and do not represent the average; getting in doesn't mean stability — the quant industry also sees layoffs, strategy decay, and performance-review pressure (actual market conditions prevail). All salary figures and acceptance ratios here are public common-knowledge ranges; defer to actual market and company conditions. Treat "preparing for the job hunt" as an investment in your own skills — even if you never enter quant, probability/statistics and programming abilities don't depreciate in the job market.