Strategy Design: From Hypothesis to Rules
Turn market ideas into testable rules without overfitting. Define entries, exits, risk, and filters for robust systems.
3.1 Market Regimes and Micro-Alphas
Regimes differ in behavior and require different edges.
Markets with clear directional movement.
Micro-Alphas
- MA crossover + pullback to EMA20
- Breakout above resistance with volume confirmation
- Higher highs/higher lows continuation
Suggested Filters
- MA slope filter (20-period slope > 0 for longs)
- ADX > 25 for trend confirmation
- ATR percentile 30–70th (avoid extremes)
Time Windows
- Avoid Asian session low volatility
- Trade London/NY overlap for momentum
3.2 Indicators and Signals: Lag, Noise, Robustness
- All indicators lag; increase smoothing → increase lag. Aim for simple, robust signals.
- Guard rails: avoid too many degrees of freedom; prefer parameter bands over exact numbers.
- Filters are safety rails, not alpha by themselves.
Relative Lag: medium
Balanced response; suitable for most strategies.
False-Signal Risk: low
Clean signals; fewer false positives but may miss opportunities.
Filter Policy Summary:
Spread Filter: ≤ 1.5× median spread ATR Filter: 20–80 percentile News Blackout: 15min before / 15min after
3.3 Algorithm Blueprint: Entries, Exits, Risk, Stops, Pauses
Provide a structured template to write rules clearly and measurably.
- Hypothesis is required
- Market regime is required
- Entry rules are required
- Exit rules are required
- Risk per trade is required
- IS/OOS split is required
3.4 Metrics Planner: Define Success/Failure Criteria
Decide what 'good enough' looks like before testing.
Expectancy (E)
0.125R
Profit Factor (PF)
1.23
Suggested Daily Risk
1.12%
Success/Failure Criteria
- PF ≥ 1.2
- Expectancy ≥ 0.05R
- Max DD ≤ 15%
- N ≥ 200 trades
- Stability: no single month worse than −10%
- E > 0 and PF ≥ 1.2 are a minimal threshold for many systems.
- Collect ≥ 200 trades for statistically meaningful evaluation.
3.5 Overfitting Guardrails
- IS/OOS split or walk-forward; never optimize on evaluation data.
- Limit parameters; prefer bands and ranges.
- Stability test: small parameter shifts change results within ±10–20%.
- Regime awareness: test across multiple market regimes.
- Include realistic costs, slippage, and news/volatility filters.
Practice
- 1.Fill the Rule Builder and Metrics Planner. Export your plan.
- 2.Define IS/OOS split and overfitting guardrails you will use.
Outcome: A one-page strategy design document with clear rules and measurable success/failure criteria.
Quiz
Lesson 3 Quiz
Test your understanding with 3 questions. Pass with 2/3 correct.
Educational content only. Not financial advice. Trading involves risk of capital loss. Past performance does not guarantee future results.