Strategy specification
Course context: This lesson belongs to a practical beginner curriculum for Pakistan.
| Component | Required rule |
|---|---|
| Universe | Exact symbols and allowed sessions |
| Regime | Trend/range/volatility definition known before entry |
| Entry | Observable trigger with no hindsight |
| Invalidation | Price/condition that proves setup wrong |
| Size | Cash risk divided by stop value per lot |
| Exit | Stop, target, time and exceptional exit |
| Portfolio controls | Maximum correlated exposure and daily drawdown |
Three candidate structures
| Structure | Hypothesis | Primary failure mode |
|---|---|---|
| Trend pullback | Continuation after controlled retracement | Range whipsaw |
| Range mean reversion | Price returns toward established value | Breakout/regime change |
| Event breakout | Volatility expands after defined catalyst/range | Slippage, false break and HMR |
| Time-based carry avoidance | Close before rollover/high-cost window | Missed continuation and extra turnover |
Validation ladder
- Freeze rules and cost assumptions.
- Backtest across multiple regimes.
- Keep an untouched out-of-sample period.
- Stress spread, slippage and gaps.
- Forward-test on demo with real timestamps.
- Calculate expectancy, drawdown and cost sensitivity.
- Run the smallest possible live pilot only if all acceptance criteria pass.
Minimum report
| Metric | Why it matters |
|---|---|
| Number of signals | Shows sample breadth |
| Expectancy in R | Normalizes position size |
| Maximum drawdown | Measures capital and psychological burden |
| Profit factor | Compares gross gains and losses |
| Longest losing sequence | Tests survival rules |
| Cost share | Shows dependence on optimistic execution |
| Out-of-sample result | Tests overfitting |
Automatic rejection rules
- Negative expectancy after complete costs.
- Most profit comes from one trade.
- Out-of-sample result materially fails.
- Small parameter changes reverse performance.
- Drawdown exceeds the pre-written maximum.
- Execution relies on fills unavailable in demo/live logs.
- The trader repeatedly breaks sizing or stop rules.
Research references
Exness Insights backtesting guide · Demo forward testing · Risk-management concepts
Questions from a first-time learner
Which CFD strategy is guaranteed to work?
None. Every strategy requires testing and can fail when conditions or costs change.
What makes a strategy testable?
Exact market, timing, entry, invalidation, size, exit, cost and stop-trading rules.
Why use out-of-sample data?
It checks whether rules were overfit to the development period.
When should a strategy be rejected?
When it fails pre-written expectancy, drawdown, robustness, execution or compliance criteria.