Developing Rule-Based Strategies From Observed Patterns

From Observation to Rule

Every systematic strategy begins with a market observation — a pattern, tendency, or structural inefficiency that appears to repeat.

The process of developing a systematic strategy is translating that observation into explicit, testable rules.

The Strategy Development Cycle

Step 1: Hypothesis Formation

Start with a specific market observation:

• "ES futures often reverses after touching the prior day's low during the London session"
• "Stocks with RSI < 30 in a broader uptrend tend to reverse within 3 days"
• "NQ breakouts above the first 30-minute range succeed 60%+ of the time in trending market conditions"

These are hypotheses — not yet tested, but specific enough to test.

Step 2: Rule Formalization

Convert the hypothesis into specific, objective rules:

Hypothesis: "ES reverses after touching the prior day's low during London"

Rules:

• Universe: ES futures
• Entry condition: Price touches (within 0.1%) the prior day's low between 3:00am and 8:00am EST, AND the current day's trend (measured by 15M VWAP slope) is upward
• Entry trigger: First 15-minute candle that closes above the prior day's low after the touch
• Stop: 8 ticks below the prior day's low
• Target: Prior day's high OR 3× initial risk, whichever comes first
• Position size: 1% of account

Step 3: Historical Testing

Apply the rules to 3+ years of historical data. Calculate the metrics from the previous backtesting lesson.

Step 4: Refinement

If the backtest shows marginal performance:

• Test small variations in the core rule (e.g., change "within 0.1%" to "within 0.2%")
• Test adding one additional filter (e.g., "only when VIX < 20")
• Test removing a rule that may be adding noise

Important: Each modification should be motivated by a logical reason, not by testing all combinations. "I'm removing this filter because it's eliminating too many valid setups without improving win rate" is a logical reason. "I tried 50 combinations and this one has the highest backtest profit" is overfitting.

Step 5: Out-of-Sample Validation

Test on the reserved out-of-sample data exactly once.

Strategy Categories and Their Characteristics

Trend-following:

• Performance: Excellent in trending markets; poor in ranges
• Characteristics: Low win rate (35–50%), high R:R ratio
• Risk: Extended drawdown periods during ranging markets
• Examples: Moving average crossovers, channel breakouts, momentum strategies

Mean-reversion:

• Performance: Excellent in ranging markets; catastrophic in strong trends
• Characteristics: High win rate (60–75%), low R:R ratio
• Risk: One trending period can wipe out many small wins
• Examples: RSI extreme reversals, Bollinger Band mean reversion, overnight gap strategies

Market-regime-dependent strategies:

• Specifically designed for one regime; include a regime filter
• Higher complexity; more robust when implemented correctly
• Examples: Volatility breakout (requires VIX above X), session fade (requires range-bound day)

Building a Strategy Portfolio

No single strategy works in all conditions. Professional systematic traders often run multiple strategies simultaneously, each suited to different regimes.

The simplest portfolio:

1. 1One trend-following strategy
2. 2One mean-reversion strategy
3. 3Market regime filter that determines which strategy is active

When the market is trending: only execute trend-following.

When the market is ranging: only execute mean-reversion.

This regime-switching reduces drawdowns significantly compared to running either strategy alone.

Strategy Development in Tradapt

Tradapt's backtesting and custom strategy features support the development cycle:

1. 1Define your strategy rules in the Custom Strategy builder
2. 2Backtest on available data
3. 3Compare backtest metrics to live journal performance
4. 4Identify gaps and refine rules

The combination of historical backtesting and live journaling in the same platform provides a complete development-to-deployment workflow.