limen.sfd.foundational_sfd

Canonical docs

In the research projects

The Python project defines its own SFD functions. The Compiled project selection and search scenarios use the declared model families wrapped here. The family reference separates those definitions from the model implementations they call.

What this package owns

Owns the production-oriented, packaged SFD modules — each defining an experiment shape via params() and a manifest() that wires the data pipeline together. Does not own model training logic (owned by the matching limen.sfd.reference_architecture model) or experiment execution.

Key entry points

Entry point	Use case	Notes
`common_strategies`	Declared comparison of the eight named baseline strategies	Rule-based inputs and execution
`dollar_bar_crash_reversal`	Crash-reversal strategy on dollar-bar features	Requires the declared dollar-bar input contract
`logreg_binary`	Standard binary-classification SFD	Manifest-driven
`dlinear_regressor`	Linear time-series regression SFD	Manifest-driven
`lightgbm_binary`	Gradient-boosted binary classifier SFD	Manifest-driven
`ridge_regressor`	Ridge next-return regression SFD	Manifest-driven
`xgboost_regressor`	Regression-style SFD	Manifest-driven
`random_binary`	Baseline classifier for comparison	Manifest-driven
`rule_based`	Predicate-driven, non-learned SFD	Uses `limen.sfd.rule_based`
`tabpfn_binary`	TabPFN-based SFD	Module is lazy; model use requires the `tabpfn` extra

Adjacent modules

limen.sfd.reference_architecture supplies the model function each foundational SFD delegates training to.
limen.experiment runs these SFDs, consuming their params() and manifest().
limen.data, limen.indicators, limen.features, limen.transforms, and limen.scalers are the building blocks the manifests wire together.

Quick orientation

Module or family	Responsibility
`tabpfn_binary.py`	Optional

text · README.md

1foundational_sfd/2├── README.md3├── __init__.py4├── _tabpfn_params.py5├── _xgboost_features.py6├── cnn1d_binary.py7├── common_strategies.py8├── dlinear_regressor.py9├── dollar_bar_crash_reversal.py10├── entropy_var_regressor.py11├── feature_link_binary.py12├── gaussian_hmm_binary.py13├── lightgbm_binary.py14├── logreg_binary.py15├── lstm_binary.py16├── pseudo_label_binary.py17├── random_binary.py18├── rbf_svm_binary.py19├── residual_regressor.py20├── ridge_regressor.py21├── rule_based.py22├── tabpfn_binary.py23├── weight_link_binary.py24└── xgboost_regressor.py

Merged source PR · 25e47866Package layout · Limen 6.0.1 · current package validation · package source

Things to know

Exports are lazy: the package uses module-level __getattr__, so importing foundational_sfd does not import heavy model dependencies until a specific SFD is accessed. This is why tabpfn_binary can be listed without tabpfn installed.
A foundational SFD owns the experiment shape; the matching reference architecture owns the training logic.
The simplest path to a new SFD is to copy an existing module, adjust params(), and modify the manifest chain.