Your personal quant research command center.
AlphaForge is a desktop application for systematic traders and quant researchers who use RealTest for backtesting. It replaces scattered spreadsheets, forgotten CSV exports, and half-remembered parameter tweaks with a structured, searchable, visual system — built to run locally on Windows with a single click.
Stop losing track of what you tested, when you tested it, and why it mattered.
Kanban board for tracking strategy statuses from inbox to deployment. Move strategies through refined, testing, paper trading, and deployed stages.
Parses RealTest CSV stats and equity curves. Archives and deduplicates .rts strategy files via SHA-256 hashing. Detects duplicate runs via parameter hashing to prevent redundant data.
Sortable metrics table comparing all ingested runs across universes. Quick-filter by strategy or status to find your best performers.
Plotly-powered interactive charts with multi-run overlays and In-Sample / Out-of-Sample split lines. Visualize drawdowns and performance metrics dynamically. A custom metrics engine and multidimensional parameter heatmaps allow for robust analysis of what matters to your specific edge.
Capture research notes with URL title fetching and link them to strategies. Never lose an idea or an observation from your research sessions.
FTS5-indexed global search accessible directly from the sidebar helps you instantly locate strategies, notes, and metrics.
Ensure your data is secure with easily accessible configuration and database backup features cleanly integrated into the application settings.
Screenshots coming soon. See Screenshot Guide for instructions on capturing them.
New to AlphaForge? Read the RealTest Workflow Guide for a step-by-step walkthrough from running a backtest to viewing the results in the dashboard.
- Python 3.11+
- Windows 10/11 (for desktop app; the dashboard runs on any OS)
- uv (recommended) or pip
git clone https://github.com/drkthng/alphaforge.git
cd alphaforge
uv sync # or: pip install -e "."alphaforge initCreates the database, data directories, and default configuration.
Copy the example config and set your RealTest output path:
cp config.yaml.example config.yamlEdit config.yaml:
paths:
realtest_output_dir: "C:/RealTest/Output" # ← Your path here# Single run with stats CSV, strategy code, and equity curve
alphaforge ingest run ./path/to/stats.csv \
--rts ./path/to/strategy.rts \
--equity-csv ./path/to/equity.csv \
--strategy "My Strategy Name"
# Bulk-scan a directory for all CSVs
alphaforge ingest scan ./path/to/outputs/ --strategy "My Strategy"alphaforge launchalphaforge build
# Output: dist/AlphaForge/AlphaForge.exeAll data stays on your machine:
data/
├── alphaforge.db # SQLite database (metadata)
├── archive/ # Versioned .rts strategy files
│ └── {strategy_slug}/
├── equity_curves/ # Parquet files (one per backtest run)
│ └── run_{id}.parquet
└── attachments/ # Uploaded images, PDFs
└── {strategy_id}/
| Command | Description |
|---|---|
alphaforge init |
Initialize database and directories |
alphaforge launch |
Start the dashboard |
alphaforge ingest run <csv> |
Ingest a single backtest run |
alphaforge ingest scan <dir> |
Scan and ingest all CSVs in a directory |
alphaforge ingest refresh |
Re-parse metrics for a run |
alphaforge build |
Build Windows desktop executable |
--rts PATH Path to .rts strategy file
--equity-csv PATH Path to equity curve CSV
--reports-dir PATH Path to HTML report directory
--strategy NAME Strategy name (auto-detected from CSV if omitted)
--universe NAME Universe name (e.g., "SP500")
--notes TEXT Notes about this run
--non-interactive Skip duplicate prompts
AlphaForge expects two types of CSV output from RealTest:
Test,Name,Dates,Periods,NetProfit,comp,ROR,MaxDD,MAR,Trades,PctWins,Expectancy,AvgWin,AvgLoss,WinLen,LossLen,ProfitFactor,Sharpe,AvgExp,MaxExp,[parameters...]
Columns after MaxExp are treated as strategy parameters.
Date,Strategy,Equity,TWEQ,Drawdown,DDBars,Daily,Weekly,Monthly,Quarterly,Yearly,M2M,MAE,MFE,Setups,Orders,Entries,Exits,Positions,Invested,Exposure
See config.yaml.example for all options.
- Folder watcher for automatic ingestion
- Remote access with authentication
- Automated cloud backup
- Walk-forward analysis UI enhancements
- Live deployment journal with broker integration
Issues, feature requests, and pull requests are welcome.
MIT



