Encyclopedia

Encyclopedia-style browse for every macroforecast schema choice. Each layer, sub-layer, axis, and option value has its own page or anchor; the tree below is generated from the live LayerImplementationSpec registry plus the OPTION_DOCS documentation registry under macroforecast/scaffold/option_docs/.

Looking for the design narrative instead? Use Architecture – that’s where the prose “why is L2 separated from L3” / “how does L7 read L4 sinks” / cross-layer reference explanations live. Encyclopedia pages here are auto-generated lookup for individual option values (description, when to use, when NOT, references, related options); Architecture pages there are hand-written narrative for the design contracts. Both are sourced from the same LayerImplementationSpec registry – encyclopedia is the machine-locked option dictionary, architecture is the human-edited design guide.

Counts

• Layers: 13

• Sub-layers: 62

• Axes (operational + future): 172

• Option values: 629

• OptionDoc entries (full prose): 611

Browse

• Browse by layer – L0 to L8 + diagnostics, table form.

• Browse by axis – every axis A-Z.

• Browse by option – every option value A-Z (e.g. ridge, pca, ar_p).

• Public Python API – curated macroforecast.run / macroforecast.replicate surface.

Layer pages

• Layer L0 – Study Setup
• Layer L1 – Data
• Layer L1.5 – Data summary
• Layer L2 – Preprocessing
• Layer L2.5 – Pre vs post preprocessing
• Layer L3 – Feature engineering
• Layer L3.5 – Feature diagnostics
• Layer L4 – Forecasting model
• Layer L4.5 – Generator diagnostics
• Layer L5 – Evaluation
• Layer L6 – Statistical tests
• Layer L7 – Interpretation / importance
• Layer L8 – Output / provenance