Portfolio R²
Track how much of a portfolio's daily variance HSR captures. Higher values indicate factor exposures are accounting for realized movement.
Hornli Specific Return
Open-Source Factor Risk Model (HSR)
Capture daily, point-in-time specific returns and factor covariance diagnostics designed for systematic investors who value transparency.
HSR is an open-source risk toolkit that powers research notebooks, production pipelines, and live trading desks. Every release is versioned, reproducible, and backed by the same methodology we run in-house.
Prefer a quick jump? Go to HSR downloads or Review the HSR FAQs.
What is a Factor Risk Model?
Factor risk models decompose portfolio variance into systematic exposures and idiosyncratic noise. By loading each security against a curated library of style, macro, and industry attributes, a practitioner can attribute performance, forecast risk, and manage exposures with far more nuance than broad beta estimates allow.
HornliQuant built HSR to make this level of rigor accessible without vendor lock-in. The model combines open data pipelines with transparent regression tooling so you can audit every step—from cross-sectional estimation to factor covariance production.
HSR at a Glance…
Whether you are stress-testing a book or prototyping a trading signal, HSR offers the coverage and cadence to keep up with your workflow. Each component is engineered for daily production use but delivered freely so you can fork, extend, and self-host.
Curious how the data evolves over time? Use the diagnostics below to inspect R², variance tracking, and asset-level coverage without leaving the page.
Portfolio-level diagnostics show how much risk the model explains and where residual variance remains. The charts pull live from the latest API payload so you can benchmark production quality in real time.
Explore trends in portfolio R², compare implied versus realized variance, and inspect the median single-stock fit without leaving the page.
Portfolio variance
Compare model-implied volatility against realized outcomes to spot over- or under-estimation of total risk.
Median single-stock R²
Gauge how well the factor set explains individual securities and whether coverage is improving across the universe.
HSR Comparison
HSR balances transparency, cadence, and cost. The table below highlights how it stacks up against typical vendor risk models.
| Capability | HSR | Typical vendor model |
|---|---|---|
| Access | Open-source repository with permissive licensing. | Closed platform with restrictive user agreements. |
| Coverage cadence | Daily updates with point-in-time history. | Often delayed, with limited history transparency. |
| Customisation | Fork, extend, or self-host the entire stack. | Limited to vendor-approved extensions. |
| Cost structure | Free download plus optional support tiers. | Multi-year contracts with user-based pricing. |
HSR FAQs
How often is the Hornli Specific Return dataset updated?
The dataset refreshes daily in mid night, and the latest timestamp is displayed in the HSR at a Glance panel above.
Can I integrate HSR outputs into my own risk engine?
Yes. The repository includes ETL scripts, regression configs, and schema definitions so you can fork the project or pipe the API outputs directly into existing infrastructure.
What licenses govern the code and datasets?
HSR source code and sample datasets are distributed under the MIT License. Attribution is appreciated but not required.
Where can I ask questions or contribute improvements?
Open a GitHub discussion or submit a pull request through the HSR repository. We review community updates regularly.