Here is the uncomfortable truth about quant portfolio projects: interviewers have seen a thousand “LSTM predicts stock prices” repos, and every one of them lowers your credibility. A project only helps if it demonstrates something the interview itself can't test — that you can handle messy data, reason about why a result might be fake, and ship working code. The projects below are ordered by difficulty, and for each one the real question is not “did you build it?” but “can you defend it under cross-examination?”
The 8 projects at a glance
| # | Project | Tier | Best for | Signal |
|---|---|---|---|---|
| 1 | Options pricer with Greeks | Beginner | Trader / QR | You understand derivatives beyond formulas |
| 2 | Pairs-trading backtest | Beginner | QR | You know what a backtest can and can't prove |
| 3 | Kelly-sizing simulator | Beginner | Trader | You think in bet sizing, not just edge |
| 4 | Portfolio optimizer with shrinkage | Intermediate | QR | You know naive Markowitz fails out of sample |
| 5 | Volatility surface fitter | Intermediate | Trader / QR | You've touched real options data |
| 6 | Walk-forward ML forecaster | Intermediate | QR / ML | You can avoid leakage on financial data |
| 7 | Limit order book simulator | Advanced | Dev / Trader | You understand matching engines & microstructure |
| 8 | Market-making bot vs. simulated flow | Advanced | Trader / Dev | You grasp inventory risk and adverse selection |
Beginner tier: prove the fundamentals
1. Options pricer with Greeks. Implement Black-Scholes, a binomial tree, and Monte Carlo, then verify they converge to each other. Compute the Greeks both analytically and by bumping inputs. The interview follow-up is predictable: why does your Monte Carlo delta disagree with the closed form, and what happens to gamma near expiry? If you can't answer, the project hurts you — the options pricing question bank covers exactly this territory.
2. Pairs-trading backtest. Cointegration test, z-score entry/exit, transaction costs included. The value is not the Sharpe ratio — it's showing you tested for lookahead bias, accounted for the fact that you selected the pair in-sample, and can explain regression to the mean without hand-waving.
3. Kelly-sizing simulator. Simulate bankroll paths under full Kelly, half Kelly, and fixed fractions for a biased coin, then extend to uncertain edge. Traders love this because it mirrors how desks actually think; the Kelly criterion guide walks through the interview versions of the same idea.
Intermediate tier: prove judgment
4. Portfolio optimizer with shrinkage. Build mean-variance optimization, then show why it explodes with a sample covariance matrix and fix it with Ledoit-Wolf shrinkage or a factor model. The before/after out-of-sample comparison is the deliverable.
5. Volatility surface fitter. Pull option chains, back out implied vols, and fit a surface with no-arbitrage sanity checks (calendar and butterfly). Cleaning real quotes — stale prices, crossed markets, wide spreads — teaches more than any textbook chapter.
6. Walk-forward ML forecaster. Predict something modest (next-day volatility, not returns) with strict walk-forward validation and purged cross-validation. The signal here is restraint: you understand why standard k-fold leaks on time-ordered data. Expect to defend every design choice with the rigor of the ML question bank and the time series bank combined.
Advanced tier: prove you can build the real thing
7. Limit order book simulator. A price-time-priority matching engine handling limit orders, cancels, and market orders, with tests. In C++ this doubles as your low-latency talking point; in Python it still demonstrates you know what happens between “send order” and “get fill.” Pair it with the market microstructure questions to speak the vocabulary fluently.
8. Market-making bot. Quote two-sided markets against simulated informed and uninformed flow, track inventory, and measure how your P&L decomposes into spread capture versus adverse selection. This is the single most interview-relevant project on the list — it is literally the game trading firms make you play.
How to present projects so they survive the interview
- One resume line, three claims: what you built, one honest quantitative result, one limitation you identified. “Found the strategy's Sharpe collapsed after costs” is a stronger bullet than a suspiciously good number.
- Negative results are fine. Interviewers trust “I tested X and it didn't work, here's why” far more than claimed alpha.
- Match project to role. Researchers should lean 2, 4, 6; traders 1, 3, 8; developers 7 plus performance work. If you're unsure which track fits you, the trader vs. researcher breakdown is the place to start.
- Two deep projects beat six shallow ones. Every line on your resume is an invitation for a 15-minute grilling — only list what you can defend.
Projects get you the interview; solving problems under time pressure gets you the offer. Drill the underlying concepts in the QuantVault problem bank, and pressure-test project #8 for real in the market-making game or the wider set of trading games.
More career guides
- Citadel Quant Salary (2026): QR, QT & Citadel Securities Comp
- A Day in the Life of a Quant Trader (Pre-Market to Close)
- Becoming a Quant Without a PhD
- Jane Street Salary (2026): Trader & Researcher Comp by Level
- Optiver Salary (2026): Trader Compensation by Office & Level
- Quant Developer Interview Guide: Role, Compensation & the Full Loop
- Quant Firms by Interview Difficulty
- The Quant Internship Timeline for Summer 2027
- The Quant Trading Interview Process: Every Stage From CV Screen to Superday
- Quant Researcher Salary (2026): New Grad to Senior, With PhD Premium Data
- How to Write a Quant Resume (2026): Templates & Real Bullet Examples
- Quant Salaries in 2026: The Honest Numbers
- Quant Trader vs Quant Researcher
- Questions to Ask at the End of a Quant Interview
- The Quant Online Assessment, Explained
- All guides & explainers
Frequently asked questions
Do quant firms actually look at portfolio projects?
Yes, but mainly as interview material rather than as a credential. A project on your resume invites detailed follow-up questions about your methodology, assumptions, and limitations. A well-defended project can carry a 30-minute interview segment, while an overstated one gets dismantled quickly.
How many quant projects should I put on my resume?
Two or three deep projects beat five or six shallow ones. Each project line is an invitation for extended questioning, so only list work you can defend in detail. Depth of understanding matters far more to interviewers than breadth of repos.
Should I claim my trading strategy is profitable?
Only if you can defend the result against lookahead bias, transaction costs, and in-sample selection effects, because interviewers will probe all three. Honest negative results are often more impressive: explaining why a strategy failed demonstrates exactly the skepticism quant firms hire for.
What is the best quant project for someone with no finance background?
An options pricer with Greeks is the standard starting point: it requires only calculus and basic probability, and it builds vocabulary you will need in interviews anyway. A Kelly-criterion bankroll simulator is a strong second choice, since it connects directly to how trading desks think about bet sizing.
Practice the real thing
QuantVault has 2,800+ quant interview problems with full solutions, intuition, and hints, firm-by-firm interview funnels, and an auto-graded coding judge. Start free.