How to Use AI Optimization in Backtest Strategy

AI optimization automatically finds the best parameters for your trading strategy. Instead of manually testing hundreds of combinations, enable one checkbox and let AI do the work.

Quick Start (3 Steps)

Step 1: Configure Basic Settings

Set only the essential parameters:

• data: Connect your OHLC data
• signals: Connect signals from Signal Generator
• initial_capital: Starting capital (e.g., 10000)
• commission: Trading commission (e.g., 0.001 = 0.1%)
• slippage: Expected slippage (e.g., 0.0005 = 0.05%)

Step 2: Enable AI Optimization

Check the "🤖 AI Find Best Strategy" checkbox.

All manual parameters (stop loss, take profit, position size) will be hidden - AI determines them automatically.

Step 3: Select Optimization Target

Choose what metric AI should optimize for:

Target	Best For
sharpe_ratio	Balanced risk-adjusted returns (recommended)
total_return	Maximum profit
profit_factor	Consistent profitability
sortino_ratio	Minimize downside risk
win_rate	Highest winning percentage

Configuration Options

SL/TP Type Selection

Choose how stop loss and take profit values are calculated:

Type	Description	Example
percent	Percentage of entry price	2.0 = 2%
fixed	Price units/pips	50 = 50 pips
atr	ATR multiples	1.5 = 1.5×ATR

Recommendation:

• Stocks/ETFs: Use percent
• Forex: Use fixed or atr
• Crypto: Use percent or atr

Top Results Count

Set ai_top_results_count to control how many results to include (1-50).

Default: 10 best parameter combinations.

Show Trades for Specific Result

Set ai_show_trades_for to see detailed trades for any top result:

• 1 = Best result (default)
• 5 = 5th best result
• 10 = 10th best result

Use this to verify calculations and compare equity curves.

Understanding AI Output

Best Parameters

Ready-to-use optimal settings:

{
  "stop_loss_value": 1.8,
  "take_profit_value": 4.5,
  "position_size": 0.15,
  "trailing_stop": true,
  "trailing_stop_value": 1.2
}

Best Results

Performance metrics for optimal configuration:

{
  "total_return_pct": 47.3,
  "sharpe_ratio": 1.85,
  "max_drawdown_pct": 12.4,
  "win_rate": 58.2,
  "profit_factor": 2.1,
  "total_trades": 156
}

Data Analysis

AI's volatility analysis of your data:

{
  "avg_bar_change_pct": 1.2,
  "volatility_95th_pct": 3.5,
  "detected_timeframe": "daily",
  "bars_analyzed": 5000
}

Recommendations

Actionable insights about your strategy:

• "✅ Strong Sharpe ratio indicates good risk-adjusted returns"
• "⚠️ Win rate below 50% - relies on large winners"
• "💡 Consider tighter trailing stop for momentum capture"

Top 10 Results

Compare alternative parameter combinations:

{
  "top_10": [
    {"rank": 1, "sharpe": 1.85, "return": 47.3, "drawdown": 12.4},
    {"rank": 2, "sharpe": 1.72, "return": 42.1, "drawdown": 10.2},
    ...
  ]
}

Recursive Search

If initial results are poor (losing money or low Sharpe), AI automatically expands the search:

Attempt	What Happens
Initial	Test baseline ranges based on volatility
Retry 1	Expand ranges 1.5x wider and tighter
Retry 2	Expand ranges 2.0x wider and tighter
Retry 3	Expand ranges 2.5x wider and tighter

Trigger conditions:

• Total return < 0%
• Sharpe ratio < 0
• Sharpe < 0.3 AND return < 2%

Output shows recursive search info:

{
  "recursive_search": {
    "performed": true,
    "depth": 2,
    "reason": "Expanded ranges to find profitable strategy",
    "improved": true
  }
}

Interpreting Sharpe Ratio

Sharpe	Rating	Meaning
< 0	Poor	Losing money or too risky
0-1	Below average	Consider revising strategy
1-2	Good	Solid risk-adjusted returns
2-3	Very good	Strong performance
> 3	Excellent	Verify not overfitted

Example Workflow

Input

{
  "data": "{{workers[1].results}}",
  "signals": "{{workers[2].signals}}",
  "initial_capital": 10000,
  "commission": 0.001,
  "ai_optimize": true,
  "ai_optimize_target": "sharpe_ratio",
  "ai_sl_tp_type": "percent"
}

Output Summary

🤖 AI analyzed 280 combinations based on your data's volatility.
Best strategy found with Sharpe ratio of 1.72.

Best Parameters:
- Stop Loss: 1.8%
- Take Profit: 4.5%
- Position Size: 15%
- Trailing Stop: Enabled (1.2%)

Performance:
- Total Return: 34.5%
- Max Drawdown: 8.9%
- Win Rate: 55.3%
- Total Trades: 187

Tips for Better Results

1. Provide Enough Data

• Minimum: 500 bars
• Recommended: 2000+ bars
• Best: 5000+ bars

2. Clean Your Signals

• Remove duplicate signals
• Use signal_mode: first in Signal Generator
• Check signal count before backtest

3. Validate with Block Analysis

After AI finds good parameters, enable:

{
  "analysis_blocks": 6
}

Check consistency score - should be 60+ for reliable strategy.

4. Compare Top Results

Don't just take #1. Compare top 5:

• Similar performance = robust
• Widely different = possibly overfitted

5. Test Different SL/TP Types

If percent doesn't work well, try atr:

{
  "ai_sl_tp_type": "atr"
}

Troubleshooting

AI Returns Poor Results

• Check if signal logic makes sense
• Try different ai_sl_tp_type
• Review data quality (gaps, outliers)
• Consider different optimization target

Too Few Trades

• Strategy may be too restrictive
• Check signal count from Signal Generator
• Consider longer data period

Inconsistent Block Performance

• Strategy may be overfitted
• Try simpler signal conditions
• Increase data sample size

---

AI optimization transforms backtesting from guesswork to science. Enable it with one click, and let the data tell you what works.