Skip to content
Roadmaps / Research and trading

Quantitative Researcher

Turns an idea about markets into a rule a computer can trade, and tests it without fooling themselves.

The job

A quantitative researcher on a systematic team turns an idea about markets into a rule a computer can trade, tests it honestly and works out what it costs to run. The team’s P&L is signal net of trading costs. The job is evidence: point-in-time data, rules fixed before the final test, an honest count of how many variants were tried, and a cost model that matches how the strategy would trade. Things go wrong through look-ahead, overfitting, and signals that are real but smaller than their costs.

Supports
Systematic trading teams
Stages
6 to learn, then proof of work
Two routes
Free resources, or SelfMadeQuant programmes and projects

The roadmap

The same stages either way. The free route uses university courses, lectures, books and papers that are free online. The SelfMadeQuant route maps each stage to our programmes and ends in projects on real market data.

  1. Stage 01 · Foundations

    Probability, statistics and Python

    Inference, regression and the Python data stack.

    • Statistical inference
    • Regression
    • pandas and NumPy

  2. Stage 02 · Core

    Time-series econometrics

    Models for data that arrives in order and does not stand still.

    • ARMA
    • Stationarity and unit roots
    • Cointegration

  3. Stage 03 · Core

    Research design and backtesting

    Point-in-time data, honest validation and a backtest you can believe.

    • Point-in-time data
    • Look-ahead and leakage
    • Deflated Sharpe ratio
    • Walk-forward testing

  4. Stage 04 · Specialist

    Statistical arbitrage

    Relative-value signals: pairs, residuals and mean reversion.

    • Pairs trading
    • Factor-residual models
    • Mean reversion

  5. Stage 05 · Specialist

    Trading costs and execution

    What a signal costs to trade, which is often more than it earns.

    • Spread and impact
    • Turnover
    • Cost-aware portfolios

  6. Stage 06 · Specialist

    Machine learning for finance

    Machine learning that survives financial data’s overlap and noise.

    • Purged cross-validation
    • Tree models
    • Feature importance

  7. Stage 07 · Proof of work

    Stress-test a factor strategy for overfitting and costs

    Use Ken French portfolios to backtest many variants of a momentum or value rule, then apply the deflated Sharpe ratio to the best one. Add a turnover-based cost estimate and answer whether the edge survives both the multiple-testing haircut and trading costs.

    Free data to start with

Practise the interview

Questions this role gets asked, each with a hint and a worked solution.

  1. You tested 300 signal variants and report the best out-of-sample Sharpe. What is missing from that claim?
  2. A macro series is dated to month-end but released two weeks later. When can it enter a signal?
  3. A signal’s predicted edge is about the same size as its estimated cost. What do you analyse before allocating capital?

Start the SelfMadeQuant route free: a free account opens one complete topic from each primer and the interview question bank. No card needed.

Start free