Factor definitions you can audit
The HSR exposure library blends public fundamentals, price dynamics, and alternative datasets with strict point-in-time
handling. Every attribute is version-controlled, with schema diffs published alongside the
open notebooks so you can trace
each change.
- Style: Beta, Volatility, Momentum, Size, Trading Activity, Growth, Earnings Yield, Value, Earnings Variability, Leverage, Dividend Yield and Management Quality flavors refreshed daily.
- Industry: GICS-based Industry binaries with survivorship-safe mappings.
- Country: One Hot encoding for each country.
Want to see how exposures impact risk? Compare factor magnitudes against the
factor covariance matrix diagnostics.
Standardization steps
- Lag fundamental items to respect filing delays and ensure fully causal exposures.
- Winsorize raw metrics at extreme quantiles (e.g., 0.1 % and 99.9 %) to suppress outliers without masking true regime shifts.
- Standardised cross-sectionally using market-capitalisation weights.
- Clips values at 3X the cap-weighted sigma.
These choices keep regressions stable even when you overlay bespoke portfolio constraints or complex optimizer objectives.
Extending the exposure deck
Hornli Quant encourages users to augment the library with bespoke signals. Because every release includes code and schema
documentation, you can append custom columns, recompute z-scores, and re-run the regression stack without breaking
compatibility with the public API or downstream analytics.
Not sure where to begin? The methodology page explains how exposures, covariance,
and specific return estimation fit together end-to-end.