How to Set Up a Backtest Correctly: Avoiding Common Pitfalls

Initial Deposit and Position Sizing

Your initial deposit in a backtest should match your actual trading capital. This sounds obvious but I see traders routinely testing with $100,000 when they plan to trade with $2,000. Why does this matter?

• Minimum order sizes: On Binance, the minimum order for BTC/USDT is ~$5. With a $100K account, that's trivial. With a $2K account and 2% risk per trade, your position size is $40 — and minimum order sizes constrain your stop-loss placement.
• Commission structure: Some exchanges have tiered fee schedules based on monthly volume. A $100K account generates more volume and qualifies for lower fees than a $2K account.
• Psychological realism: A 15% drawdown on $100K is $15,000. On $2K it's $300. The strategy might survive both, but will you?

Position Sizing Methods Comparison

Different position sizing methods produce dramatically different risk profiles. Here's how they compare in practice:

For most traders, fixed percentage (1-3% per trade) provides the best balance between growth potential and capital preservation. A recent backtest of a breakout strategy on EUR/USD showed 2% risk per trade generated 127% returns over 3 years with 18% maximum drawdown, while 5% risk achieved 284% returns but suffered 41% drawdown.

Commission and Fee Modeling

This is where most backtests go wrong. Here's a cheat sheet for accurate fee modeling:

Pro tip: when in doubt, double the fees. Better to underestimate your strategy's performance than to overestimate it. If your strategy is profitable with 2× fees, it'll almost certainly be profitable with real fees.

Real-World Commission Impact: Case Study

Consider a scalping strategy tested on ETH/USDT with these parameters:

• Average trade duration: 15 minutes
• Target profit: 0.3% per trade
• Stop loss: 0.2%
• Win rate: 62%
• Trades per day: 8

Backtest results with different commission assumptions:

This demonstrates why accurate fee modeling is crucial — the same strategy goes from highly profitable to unprofitable based solely on cost assumptions.

The Hidden Cost: Funding Rates

If you're backtesting perpetual futures strategies, funding rates are a cost that most beginners overlook entirely. On Binance and Bybit, funding is charged every 8 hours — typically around 0.01% per interval during neutral markets. That sounds trivial, but it compounds: 0.01% × 3 daily × 365 days = 10.95% annually. For a swing trading strategy that holds positions for days, funding can consume a significant portion of profits.

The situation worsens during trending markets. In strong bull markets, funding rates for long positions routinely spike to 0.05–0.1% per 8 hours. A 5-day hold at 0.05% funding costs 0.75% — which might exceed your entire take-profit target on lower timeframes. Always include funding rate modeling in futures backtests. StratBase.ai automatically incorporates historical funding rate data for supported instruments, giving you realistic cost projections.

Historical Funding Rate Analysis

Analysis of BTC perpetual funding rates from 2021-2024 reveals distinct patterns:

A momentum strategy holding long positions for 5-7 days would have paid approximately 8.2% annually in funding costs during the 2021 bull run, versus receiving 1.8% during the 2022 bear market. These costs can transform a profitable strategy into a losing one.

Data Period Selection

The testing period must include different market conditions. Here's a minimum standard:

For crypto: 2020–2025 captures a complete cycle — COVID crash (March 2020), bull run (2020–2021), bear market (2022), recovery and new highs (2023–2025). Three years minimum, five years recommended.

For forex: 5–10 years. Forex cycles are longer than crypto. You need to capture multiple central bank policy cycles, at least one risk-off event, and both trending and ranging periods.

For stocks: 5–10 years minimum. Must include a significant drawdown (2020 COVID, 2022 rate hikes) and a bull market (2023–2025 AI rally).

Market Regime Analysis: Why Period Selection Matters

Testing the same RSI divergence strategy across different market periods reveals how dramatically performance varies:

A strategy tested only on 2020-2021 data would show exceptional results (73% win rate, 67% annual return) but fail miserably in the 2022 bear market. This is why comprehensive period selection is non-negotiable.

Slippage Configuration

Slippage is the difference between the price you expect and the price you actually get. It's almost always negative — you get filled worse than expected.

Conservative slippage settings:

• Crypto spot (major pairs): 0.05% per trade
• Crypto futures: 0.03–0.05%
• Forex majors: 0.2–0.5 pips
• US large-cap stocks: $0.01–0.03 per share
• US mid/small-cap stocks: $0.05–0.20 per share

Double these values for strategies that enter during high-volatility periods (breakouts, news events, market open).

Measuring Real Slippage: Exchange Analysis

Actual slippage data from live trading across different exchanges and conditions:

These measurements come from analysis of 10,000+ actual trades across different market conditions. Note how slippage increases dramatically during volatile periods and with larger order sizes.

Timeframe Selection and Its Impact on Results

Your choice of timeframe doesn't just affect signal frequency — it fundamentally changes the character of your strategy. The same RSI-based system will produce completely different results on 15-minute versus 4-hour charts, and understanding why helps you choose the right timeframe before you run a single test.

Lower timeframes (1m–15m) generate more signals but each signal carries less statistical weight. Noise dominates at these levels, so strategies need very tight stops and fast exits. Commission costs consume a larger percentage of each trade's profit. A 0.1% round-trip fee against a 0.3% target means you need to be right significantly more often just to break even.

Higher timeframes (4H–daily) produce fewer but more reliable signals. Each candle represents hours of aggregated market activity, filtering out much of the intraday noise. Stops are wider in absolute terms but smaller relative to the expected move. The trade-off is patience — you might wait days or weeks between signals.

A practical approach: start with 4H for swing strategies or 1H for intraday, then only move to lower timeframes if you have a specific reason and can demonstrate profitability after realistic fees.

Timeframe Performance Analysis

The same breakout strategy tested across different timeframes shows how dramatically results change:

Notice how lower timeframes generate more trades but worse results due to noise and higher transaction costs, while higher timeframes produce fewer but more profitable signals.

Step-by-Step Backtest Setup Process

Follow this systematic approach to ensure your backtest setup is bulletproof:

Step 1: Define Your Trading Parameters

• Determine your actual trading capital (not aspirational)
• Choose your risk per trade (1-3% recommended)
• Select instruments you'll actually trade
• Pick timeframes matching your availability

Step 2: Configure Market Data

• Set start date to include complete market cycles
• Verify data quality (check for gaps, splits, dividends)
• Include dividends for stock strategies
• Adjust for splits and corporate actions

Step 3: Set Realistic Costs

• Input your exact broker commission rates
• Add conservative slippage estimates
• Include funding costs for futures/CFDs
• Account for spread costs in forex

Step 4: Position Sizing Configuration

• Use percentage-based position sizing
• Set maximum position size (5-10% of capital)
• Configure minimum order size constraints
• Enable reinvestment of profits

Step 5: Strategy Rules Implementation

• Code entry conditions precisely
• Define exit rules (profit targets, stops, time)
• Eliminate subjective decision points
• Test with and without filters

Step 6: Validation and Sensitivity Analysis

• Run backtest on full dataset
• Test with 2x commission rates
• Vary position sizing (1%, 2%, 3%)
• Check results on different time periods

Advanced Configuration: Walk-Forward Analysis

Static backtests assume your strategy parameters remain optimal throughout the entire testing period. Walk-forward analysis provides a more realistic assessment by periodically re-optimizing parameters and testing forward performance.

Implementation process:

1. Anchor period: Use first 12 months to optimize parameters
2. Out-of-sample test: Trade next 3 months with optimized settings
3. Re-optimization: Add the 3-month period to data and re-optimize
4. Repeat: Continue rolling forward through entire dataset

Example results from EUR/USD momentum strategy:

Walk-forward analysis typically reduces apparent returns by 20-40% but provides more realistic expectations for live trading performance.

“The backtest is not your friend — it's your quality control department. Give it every reason to reject your strategy. The strategies that survive harsh testing are the ones worth trading.”

Checklist Before Running Your Backtest

Run through this checklist every time:

1. Initial deposit matches your real capital
2. Commissions set to your exchange's actual rates (or higher)
3. Slippage enabled (minimum 0.05% for crypto)
4. Data period includes at least one bull and one bear market
5. Position sizing uses percentage risk, not fixed size
6. Stop-loss and take-profit are defined in the rules
7. Strategy rules are fully mechanical — no subjective decisions
8. Funding rates included for perpetual futures strategies
9. At least 30 trades expected for statistical significance
10. Out-of-sample period reserved for final validation
11. Multiple timeframes tested for robustness
12. Worst-case cost scenarios modeled

For a comprehensive overview of the entire backtesting process, see the complete backtesting guide. Understanding why strategies fail will also help you avoid the most common setup mistakes.

FAQ

Further Reading

• RSI on Investopedia
• Backtesting on Investopedia
• Sharpe Ratio on Investopedia