Strategy Optimization

Optimization tests multiple parameter combinations to find the best-performing settings for your strategy. Available for Pro+ subscribers.

1. Select parameters to optimize (indicator periods, thresholds, SL/TP values)
2. Define ranges for each parameter: min, max, and step
3. System generates combinations — all possible value sets (Cartesian product)
4. Each combination runs a full backtest
5. Results are sorted by your chosen objective metric

Example

Optimize RSI period and SL percentage:

• RSI period: min=10, max=20, step=2 → values: 10, 12, 14, 16, 18, 20 (6 values)
• SL: min=1%, max=5%, step=1% → values: 1, 2, 3, 4, 5 (5 values)
• Total combinations: 6 × 5 = 30 backtests

| Plan | Parameters | Max Combinations | |------|-----------|-----------------| | Free | — | Not available | | Pro | 1 parameter | Up to 10,000 | | Premium | Unlimited | Up to 10,000 | | Private | Unlimited | Up to 10,000 |

Sort results by different metrics to find the best combination:

| Objective | Best For | |-----------|----------| | Net P&L | Maximize absolute profit | | Win Rate | Maximize percentage of winning trades | | Sharpe Ratio | Maximize risk-adjusted returns | | Profit Factor | Maximize profit/loss ratio |

The Optimization tab shows a table of all tested combinations:

• Each row = one parameter combination
• Columns show the parameter values and resulting metrics
• Sorted by your chosen objective (best first)
• Click any row to view the full backtest results for that combination

Avoid Overfitting

• Don't optimize everything — optimize 1-2 key parameters, not all
• Use longer test periods — more data = more robust results
• Watch for outliers — if the best result is far better than the rest, it may be an anomaly
• Out-of-sample testing — run the optimized parameters on a different time period

Parameter Selection

• Start with the most impactful parameters (SL/TP, main indicator period)
• Use reasonable ranges (don't test RSI period 2 to 500)
• Use appropriate step sizes (too fine = too many combos, too coarse = miss optimal)

Tip: If all combinations in a narrow range perform similarly well, the strategy is robust. If only one specific set works, the strategy is likely overfitted.

Walk-Forward testing is planned for a future release. It will:

1. Optimize on a training period
2. Test on an out-of-sample period
3. Roll forward and repeat
4. Aggregate results to assess robustness

Q: How long does optimization take? A: Each combination runs a full backtest (5-30 sec). 100 combinations ≈ 8-50 minutes. The system processes them in parallel where possible.

Q: Can I cancel an optimization? A: Yes. Cancel from the backtest page. Partial results up to the cancellation point are saved.

Q: Is the best optimization result always the best strategy? A: Not necessarily. The "best" in-sample result may be overfitted. Compare top results — if they cluster in a similar parameter range, confidence is higher.

• Key Metrics Explained
• Understanding Backtest Results
• StratBase Score & Risk Score