# Strategy Generator Studio

Strategy Generator Studio is a Python research tool for discovering, backtesting, ranking, and exporting rule-based long-only trading strategies.

It supports:

- GUI-driven research runs with `PyQt5`
- Random search and grid search strategy discovery
- Rule-based signal compilation from human-readable expressions
- ATR stop-loss / take-profit / trailing-stop backtests
- Multi-horizon weekly consistency ranking
- Optional credibility screening and three-phase validation
- Exportable HTML dashboards and standalone strategy scripts

## Quick Start

1. Install Python 3.10+.
2. Install dependencies:

```bash
pip install pandas numpy matplotlib yfinance PyQt5 TA-Lib
```

3. Launch the GUI:

```bash
python strategy_gui.py
```

4. Or run a scripted example:

```bash
python example_random_discovery.py
```

## Main Files

- `strategy_gui.py`: desktop app and run controls
- `run_backtest_report.py`: research engine, ranking, validation, exports, dashboards
- `backtest.py`: long-only backtest execution
- `rules.py`: rule parser/compiler
- `indicators.py`: indicator cache backed by TA-Lib
- `conditions.py`: implemented entry/exit condition registry
- `combos.py`: strategy-combination generation

## Outputs

Each run creates a timestamped folder under `reports/`, usually containing:

- `strategies_report.csv`
- `strategies_report_with_meta.csv`
- `credible_candidates.csv` or `ranked_candidates.csv`
- `screened_out_candidates.csv`
- `fallback_ranked_candidates.csv` when nothing passes the screen
- `dashboards/index.html`
- `dashboards/top_ranked_table.csv`
- per-strategy dashboard HTML files
- exported standalone strategy scripts in `reports/output/<run_name>/`

## Read Next

- User guide: [docs/USER_GUIDE.md](docs/USER_GUIDE.md)
- Code manual: [docs/CODE_MANUAL.md](docs/CODE_MANUAL.md)