Research you can inspect
From a data panel to a portfolio report
AlphaForge is a Python research toolkit connecting factor evaluation, walk-forward prediction, constrained portfolios, execution costs and attribution. The CLI, FastAPI service and Streamlit dashboard use the same research pipeline.
Synthetic data, computed results
The published example uses the bundled synthetic provider. Its metrics demonstrate the software and the effects of its assumptions. They are not evidence of a market anomaly or an investable strategy.
Explore the system
| Start here | What you can inspect |
|---|---|
| Sample run | Actual metrics, daily returns, configuration, dependency versions and source hashes |
| Quickstart | Local installation, one-command research run, API and dashboard |
| Architecture | The stages and shared state connecting data to reports |
| Validation | Automated checks, interpretation limits and remaining work |
A complete research path
flowchart LR
A[Data + quality] --> B[42 factors]
B --> C[Walk-forward model]
C --> D[Constrained portfolio]
D --> E[Execution + costs]
E --> F[Attribution + report]- Read the data assumptions. Data & Quality describes provider behavior and survivorship limits.
- Understand the signal. Factor Research and Portfolio Optimization explain the measurements and constraints.
- Follow the cash. Backtesting describes execution timing, cash accounting and modeled costs.
- Inspect explanations. The Research Copilot defaults to deterministic summaries of pipeline outputs, with no paid API required.
Reproduce locally
git clone https://github.com/dev-belly/alphaforge.git
cd alphaforge
python -m venv .venv
source .venv/bin/activate
pip install -e ".[viz]"
python scripts/publish_sample.py
The example opens directly from docs/sample/research_report.html.
See the recorded environment and outputs before comparing numbers.