Backtesting: Testing the Future on the Past

Backtest simulates bot performance on historical price data. Lets you see what would have happened if bot traded before.

Why Backtests Matter?

Without backtest:

• Launch bot blindly
• Rely on intuition
• Risk losing money

With backtest:

• See potential results
• Adjust parameters
• Launch only proven strategy

How Backtesting Works in AI Traders?

Step 1: Create Configuration

1. Click Backtest
2. Select:
   • Trading pair (BTC/USDT)
   • Strategy (DCA, grid, martingale)
   • Parameters (amount, intervals, filters)

Step 2: Choose History Period

Minimum 3 months (see seasons)
Maximum 5 years (full cycle)

Recommend: 6–12 months ago.

Step 3: Run Test

Bot simulates trading on historical data. Takes 30 seconds to 5 minutes.

Step 4: Read Results

Key metrics:

Metric	Meaning
Total Profit	How much you'd earn
ROI %	Percent on capital
Max Drawdown	Biggest balance drop
Win Rate	% of winning trades
Trade Count	How many trades

Example Backtest: DCA on BTC

Setup:

• Period: 2025-01-01 to 2026-01-01
• Strategy: DCA
• Investment: $100 weekly
• Filter: RSI < 30

Results:

Total Investment: $5,200 (52 weeks × $100)
Current Position Value: $7,850

💰 Profit: +$2,650 (+51% ROI)

Max Drawdown: -18% (when BTC fell 20%)
Win Rate: 78% (39 weeks profit, 13 loss)
Trades: 52 (one per week)

Conclusion: strategy worked well! Can launch on real money.

Interpreting Results

Red Flag 🚩: ROI < 0%

Result: -$500 loss
Conclusion: strategy LOST in this period
Action: ✅ CHANGE PARAMETERS or choose different pair

Yellow Flag 🟡: Max Drawdown > 40%

Max Drawdown: -45%
Conclusion: your balance dropped in half
Risk: you might panic
Action: ⚠️ REDUCE investment sizes

Green Flag 🟢: ROI > 20%, Drawdown < 30%

Result: +$1,000 profit
Max Drawdown: -20%
Conclusion: good strategy with acceptable risk
Action: ✅ CAN LAUNCH

Backtesting Mistakes

Mistake 1: Overfitting

❌ Wrong:

Test 100 parameter combinations
One gives +100% profit
Choose that one
On real money: -50% loss

Why? Parameters perfectly fitted to past, future is different.

✅ Right:

Choose reasonable parameters (theory-based)
Test on 3 different periods
All show +15–30% ROI
Choose that strategy

Mistake 2: Ignore Max Drawdown

❌ Wrong: see +100% ROI, launch, see -60% drawdown, panic, stop

• Strategy would have recovered
• But you quit too early

✅ Right: choose strategy with drawdown you can handle

Mistake 3: One Period Isn't Enough

❌ Wrong: test only uptrend period (Jan 2025)

• Get +200% ROI
• Launch, market crashes
• Strategy doesn't work on downtrend

✅ Right: test minimum 6–12 months (different scenarios)

Pre-Launch Checklist

• Backtest shows ROI > 0%
• Max drawdown < 40% (you can handle psychologically)
• Win rate > 50% (more wins than losses)
• Tested on 3+ different periods
• Parameters are reasonable (not overfitted)
• Starting with small amount (10% of capital)

Next Step

1. Run first backtest on DCA BTC 6 months ago
2. Check results — should be positive
3. Try changing parameters — see how results change?
4. Launch on real in Paper Trading mode
5. Compare reality vs backtest — usually close but not perfect

Remember: backtest is a guide, not a guarantee. Real market can surprise!