RizonA quant project for your resume: build a market maker on a live order book
The project that shows what trading firms test for. Build a market maker, run it on a live book with paper money, and measure where its money went.
Short answer: Build a market maker and run it on a live order book, then measure where its money went. It's one of the few projects that shows what trading firms actually test: pricing under uncertainty, inventory, adverse selection and keeping a system running. A backtest can't teach that, because in a backtest nobody hits your quotes. On Rizon you can do it for free: a key takes one call, the book is real, the money is paper, and your account sits on a public leaderboard you can link to.
Why this project
Most quant projects on resumes look alike: a moving-average backtest, an options pricer, an LSTM on stock prices. They're fine as practice, and they're hard to talk about for more than a minute. A market maker gives you things to talk about, because it fails in specific, measurable ways.
| What it shows | Where it comes up |
|---|---|
| Pricing under uncertainty | Where is fair, and how sure are you? Every quote is an answer |
| Adverse selection | Why your fills lose money right after they happen |
| Inventory risk | What you do when you've bought too much and the price keeps falling |
| Expected value, in your own data | The spread you capture against what your fills lose |
| Engineering | Reconnects, rate limits, cancels on shutdown, logs you can trust |
Firms do look for this. Tower Research writes: "Lots of applicants say they know Python. Fewer can point to projects where they shipped something: a backtester, a market simulator, an analytics tool." It asks for "clear hypotheses, thoughtful evaluation, honest discussion of limitations, and reproducible code."
It's also the shape of the interview. Trading firms run market making games for trader roles, where you quote a two-sided price and defend it. Having run a real one changes how you answer.
What to build, in four steps
| Step | Build | What you should be able to say after |
|---|---|---|
| 1 | Post a bid and an ask around fair on one book, and cancel them on shutdown | How a limit order book works, and what it means to rest and to take |
| 2 | Re-quote as the price moves, inside the rate limit | Why a slow quote costs money. A bot here picks off stale ones, on purpose |
| 3 | Skew your quotes against your inventory | How much you widen or shift, and why that number |
| 4 | Log every fill and mark it to the price 1 and 5 seconds later | Your realized spread: what you earned, net of what the flow knew |
Step 4 is what separates this from a demo. Most people report PnL. Reporting the spread you captured and the part adverse selection took back is what an interviewer will want to dig into.
Getting started
The whole stack is an API key and a few hundred lines. You can write it yourself, which is better for an interview, or start from working code and take it apart.
| Start from | What it is |
|---|---|
| The API docs | REST and a WebSocket, with a key in one call: curl -X POST https://rizon.markets/api/v1/accounts |
| A starter repo | A small market maker and a tested prompt, if you want Claude Code to draft it |
| An open-source market maker | The bots that quote Rizon's own books: distributions, ladders, guards, and tests. Good for reading, too big to submit as your own |
If an AI wrote any of it, say so, and make sure you can defend every line. Interviewers can tell within two questions.
What goes on the resume
Numbers and a link, not adjectives:
Built a market maker in Python on a live order book: 40,000 fills over 30 days. Measured realized spread at 1 s and 5 s, and cut adverse selection by 35% by widening on volatility and skewing on inventory. Code and public record linked.
Those numbers are an example of the form. Use your own, even if they're bad: "my first version lost money and here's why" is a better interview than a suspiciously smooth curve.
What it won't show
This project shows what you built. For the interview round itself, practice quoting with make me a market.
Be straight about the limits, because you'll be asked. It's paper money, so nothing you did was tested by fear. Most of the flow on Rizon comes from bots, so your adverse selection will be lower than on a real venue. And one project doesn't replace the fundamentals: firms still test probability, mental math and coding directly.
FAQ
What's the best quant project for a resume?
One you can defend in depth, that has real data and a measured result. A market maker fits that well because it forces you to deal with fills, inventory and adverse selection, which backtests hide.
Do I need a finance background to be a quant?
No. Jane Street says "Finance background: optional", and Optiver says "You don't need experience in trading or finance to get started." What they test is thinking: probability, problem solving and how you reason out loud.
Should I put a backtest on my resume?
It's fine as a line, but be ready for "what would you expect to change live?" A backtest assumes you get filled at prices you'd really have missed. A live project answers that question for you.
Python or C++?
Python is fine for a project at this speed, and easier to show. Trading firms use both. The numbers you measured will matter more than the language.
Can I use a paper exchange for a real project?
Yes, if you're honest that it's paper. What the project shows is your reasoning and your measurement, and those don't depend on the money being real.