Tier 6 · Master · Module 6.1
Quant / Systematic Trading
Systematic trading as a specialisation — fully rule-based strategies, research workflow, robustness testing, walk-forward validation, and running a portfolio of uncorrelated systems.
Lesson 4 of 5 · 5 min read
Systematic traders turn every decision into a rule, test those rules on data, and let the system execute — often automatically. The work shifts from watching charts to research: forming hypotheses, testing them honestly, and managing a portfolio of strategies. It builds directly on backtesting (Module 3.4) and automation (Module 4.3), and it rewards scientific discipline more than market intuition.
What you'll learn
- What systematic trading involves, and how it differs from discretionary trading
- The research workflow: hypothesis to live system
- Robustness testing: parameter stability, out-of-sample, and walk-forward
- Running a portfolio of systems
- The skills and temperament it requires
1. What it is
| Discretionary | Systematic | |
|---|---|---|
| Decisions | Judgement within a plan | Fully rule-based |
| Testing | Partial (manual backtests) | Complete — every rule is testable |
| Execution | Manual | Often automated |
| Main risk | Inconsistent execution | Overfitting and model decay |
| Daily work | Analysing charts | Research, monitoring, maintenance |
2. The research workflow
- Hypothesis — an idea with a logical reason to work (for example, "trends in major currency pairs tend to persist over weeks because macro drivers change slowly").
- Specification — exact rules, decided before testing.
- In-sample test — does the idea show an edge after costs?
- Robustness tests — does it survive variations and unseen data?
- Forward test — does it work in real time, with real execution?
- Live, small — scale up under clear rules.
- Monitor — track live results against expectations; retire systems that decay.
3. Robustness testing
A robust strategy works across reasonable variations — not only at one lucky setting.
| Test | What it checks |
|---|---|
| Parameter stability | Do nearby settings produce similar results? (Look for a plateau, not a spike — see Intro to automated backtesting concepts) |
| Out-of-sample | Does it work on data not used in development? |
| Walk-forward analysis | Repeatedly optimise on one window, then test on the next unseen window, rolling through history |
| Multiple markets | Does the logic work on related markets? |
| Cost stress | Is it still profitable with costs 50–100% higher than expected? |
| Monte Carlo | Reshuffling the trade sequence many times: how bad could drawdowns realistically get? |
Worked example: walk-forward
(Illustrative.) Ten years of data.
| Window | Optimise on | Test on |
|---|---|---|
| 1 | Years 1–3 | Year 4 |
| 2 | Years 2–4 | Year 5 |
| … | … | … |
| 7 | Years 7–9 | Year 10 |
Joining the seven test years gives a performance record made entirely of results the optimisation never saw — a far more honest estimate than a single optimised backtest.
4. A portfolio of systems
Systematic traders often run several strategies whose returns aren't closely correlated — for example, a trend-following system and a mean-reversion system on different markets. When one is in drawdown, the other may not be.
- Allocate risk across systems, not just across trades.
- Monitor each system's rolling expectancy against its benchmark (see Full journal with equity curve, drawdown, and expectancy tracking).
- Set retirement rules: for example, suspend a system if its drawdown exceeds 1.5 × its worst out-of-sample drawdown.
5. Skills and temperament
| May suit you if… | May not if… |
|---|---|
| You enjoy data, testing, and (ideally) coding | You prefer reading charts in real time |
| You can accept rules overriding your opinion | You'd override the system when it "looks wrong" |
| You're patient with research that often finds nothing | You want quick results |
| You think in probabilities and samples | You judge by the last few trades |
Common beginner mistakes
- Optimising first, thinking later.
- Testing hundreds of variations and keeping the best (data snooping).
- Skipping cost stress tests.
- Overriding the system during drawdowns.
- Never retiring a system that has clearly decayed.
Key terms
| Term | Meaning |
|---|---|
| Systematic trading | Fully rule-based trading, usually tested and automated |
| Hypothesis | A testable idea with a logical rationale |
| Walk-forward analysis | Rolling optimisation and out-of-sample testing through history |
| Monte Carlo analysis | Reshuffling or resampling trades to estimate the range of outcomes |
| Model decay | A strategy's edge weakening over time |
| Retirement rule | A pre-set condition for suspending a system |
Practice
- Write one hypothesis with a logical reason it should work, and a full test plan — before testing.
- Run the in-sample test, then a parameter-stability check on nearby settings.
- Run an out-of-sample or simple walk-forward test.
- Stress-test costs at +50%. Does the edge survive?
Quick recap
- Systematic trading makes every decision a rule, tested before use.
- Follow the workflow: hypothesis → specification → test → robustness → forward → live → monitor.
- Demand robustness: plateaus, out-of-sample, walk-forward, cost stress.
- Run a portfolio of uncorrelated systems with retirement rules.
- It suits traders who value evidence over opinion.
Educational content only — not financial advice. Trading involves substantial risk of loss. Practise on a demo account before risking real money.
