5.1. Data Quality and Costs: What Must Be Realistic

Garbage in → garbage out. Ensure sufficient, continuous history; set spreads, commissions, swaps, and slippage realistically.

Compounding (reinvestment) changes equity shape; align with your live plan.

Data Quality Auditor

History coverage: IS ≥ 24 months; OOS ≥ 6 months.

More history captures different market regimes.

Continuity: no long gaps; reload history if needed.

Gaps can cause false signals or missed entries.

Tick vs 1-min OHLC: match your logic's granularity.

Scalpers need ticks; swing traders can use OHLC.

Broker settings: correct contract size, min stops, margin.

Mismatched settings = wrong position sizing results.

Time zone alignment: news/rollover awareness.

Rollover spreads widen; news times vary by broker TZ.

Cost Modeler (Expectancy Impact)

Spread (pips): 1

Commission ($/lot RT): 7

Swap ($/lot/day): -1.2

Slippage (pips): 0.2

Stop distance (pips): 30

Win rate %: 45

Avg win (R): 1.5

Avg loss (R): 1

Results

Cost per trade (pips):1.90

Cost per trade (R):0.063

Raw Expectancy (R):0.125

Net Expectancy (R):0.062

Compounding ON

Compounding doesn't create edge; it magnifies it.

5.2. Statistical Traps: Bias Lab

Most backtest failures stem from hidden biases. Name them, prevent them, and document safeguards.

Look-Ahead Bias

Data-Mining Bias

Survivorship Bias

Period Fitting

5.3. IS/OOS Splits and Multi-Scenario Planning

Separate optimization (In-Sample) from evaluation (Out-Of-Sample). Test across symbols/timeframes to avoid regime dependence.

IS/OOS Split Planner

Aim IS ≥ 24m, OOS ≥ 6m.

Next-bar execution

Split: 70/30 | Execution: Next-bar

Multi-Test Matrix

5.4. Practice: Rerun with Realistic Costs and an OOS Window

Step A: Configure realistic costs

Set spread = 1.0 pip, commission = $7/lot, slippage = 0.2 pips, swap = −$1.2/lot/day.

Ensure 'Use Date' for both IS and OOS windows.

Step B: Run IS and OOS

Run optimization on IS only; freeze parameters.

Run a single pass on OOS; do not re-optimize.

Step C: Compare and export

Compare PF, Expectancy, Max DD. Note stability across months.

Export HTML + CSV reports; name files clearly (symbol_tf_is/oos_modeling_costs).

Sample Reports

Open Sample Report A (optimistic)Open Sample Report B (realistic)

5.5. Report Comparator (A/B Quick View)

Spot the difference: costs, slippage, and IS/OOS separation often turn 'amazing' systems into average ones — that's truth.

Highlight differences > 20%

Metric	Report A (Optimistic)	Report B (Realistic Costs)
Net Profit	$12,450	$4,320
Profit Factor	2.85	1.42
Max Drawdown %	8.2%	15.7%
Total Trades	847	821
Expectancy (R)	+0.45	+0.12
Win Rate	68%	52%

5.1. Data Quality and Costs: What Must Be Realistic

Garbage in → garbage out. Ensure sufficient, continuous history; set spreads, commissions, swaps, and slippage realistically.

Compounding (reinvestment) changes equity shape; align with your live plan.

Data Quality Auditor

History coverage: IS ≥ 24 months; OOS ≥ 6 months.

More history captures different market regimes.

Continuity: no long gaps; reload history if needed.

Gaps can cause false signals or missed entries.

Tick vs 1-min OHLC: match your logic's granularity.

Scalpers need ticks; swing traders can use OHLC.

Broker settings: correct contract size, min stops, margin.

Mismatched settings = wrong position sizing results.

Time zone alignment: news/rollover awareness.

Rollover spreads widen; news times vary by broker TZ.

Cost Modeler (Expectancy Impact)

Spread (pips): 1

Commission ($/lot RT): 7

Swap ($/lot/day): -1.2

Slippage (pips): 0.2

Stop distance (pips): 30

Win rate %: 45

Avg win (R): 1.5

Avg loss (R): 1

Results

Cost per trade (pips):1.90

Cost per trade (R):0.063

Raw Expectancy (R):0.125

Net Expectancy (R):0.062

Compounding ON

Compounding doesn't create edge; it magnifies it.

5.2. Statistical Traps: Bias Lab

Most backtest failures stem from hidden biases. Name them, prevent them, and document safeguards.

Look-Ahead Bias

Data-Mining Bias

Survivorship Bias

Period Fitting

5.3. IS/OOS Splits and Multi-Scenario Planning

Separate optimization (In-Sample) from evaluation (Out-Of-Sample). Test across symbols/timeframes to avoid regime dependence.

IS/OOS Split Planner

Aim IS ≥ 24m, OOS ≥ 6m.

Next-bar execution

Split: 70/30 | Execution: Next-bar

Multi-Test Matrix

5.4. Practice: Rerun with Realistic Costs and an OOS Window

Step A: Configure realistic costs

Set spread = 1.0 pip, commission = $7/lot, slippage = 0.2 pips, swap = −$1.2/lot/day.

Ensure 'Use Date' for both IS and OOS windows.

Step B: Run IS and OOS

Run optimization on IS only; freeze parameters.

Run a single pass on OOS; do not re-optimize.

Step C: Compare and export

Compare PF, Expectancy, Max DD. Note stability across months.

Export HTML + CSV reports; name files clearly (symbol_tf_is/oos_modeling_costs).

Sample Reports

Open Sample Report A (optimistic)Open Sample Report B (realistic)

5.5. Report Comparator (A/B Quick View)

Spot the difference: costs, slippage, and IS/OOS separation often turn 'amazing' systems into average ones — that's truth.

Highlight differences > 20%

Metric	Report A (Optimistic)	Report B (Realistic Costs)
Net Profit	$12,450	$4,320
Profit Factor	2.85	1.42
Max Drawdown %	8.2%	15.7%
Total Trades	847	821
Expectancy (R)	+0.45	+0.12
Win Rate	68%	52%

Truthful Backtests in MT5 — How Not to Fool Yourself

5.1. Data Quality and Costs: What Must Be Realistic

Results

5.2. Statistical Traps: Bias Lab

5.3. IS/OOS Splits and Multi-Scenario Planning

5.4. Practice: Rerun with Realistic Costs and an OOS Window

Step A: Configure realistic costs

Step B: Run IS and OOS

Step C: Compare and export

Sample Reports

5.5. Report Comparator (A/B Quick View)

Lesson 5 Quiz

Truthful Backtests in MT5 — How Not to Fool Yourself

5.1. Data Quality and Costs: What Must Be Realistic

Results

5.2. Statistical Traps: Bias Lab

5.3. IS/OOS Splits and Multi-Scenario Planning

5.4. Practice: Rerun with Realistic Costs and an OOS Window

Step A: Configure realistic costs

Step B: Run IS and OOS

Step C: Compare and export

Sample Reports

5.5. Report Comparator (A/B Quick View)

Lesson 5 Quiz