ai-asset-pricing: Skill for Claude Code

Results Reporter

Automatically generates a structured report folder after every PyBondLab run.

When to Apply

After any call to StrategyFormation.fit(), BatchStrategyFormation.fit(), BatchWithinFirmSortFormation.fit(), or DataUncertaintyAnalysis.fit() — always invoke the reporter before presenting results.

Usage

from PyBondLab.report import ResultsReporter

reporter = ResultsReporter(
    result=result,          # FormationResults or BatchResults
    mnemonic='cs_single_5', # short name
    script_text=SCRIPT,     # the Python code that produced result
    output_dir='results',   # root directory (or project scripts/tests/{test}/output/)
)
report_path = reporter.generate()

Mnemonic Convention

Strategy	Pattern	Example
SingleSort	{signal}_single_{nport}	cs_single_5
DoubleSort	{var1}_{var2}_double_{n1}x{n2}	rat_cs_double_3x5
WithinFirmSort	{signal}_wfs	cs_wfs
Batch SingleSort	batch_{n_signals}s	batch_3s
Batch WithinFirm	batchwfs_{n_signals}s	batchwfs_3s
FF-style	ff_{var1}_{var2}_{n1}x{n2}	ff_sze_bbtm_2x3

Output Structure

Single strategy: results/{mnemonic}_{YYYY_mm_dd}/ with meta.json, script.py, tables/summary_stats.csv, figures/portfolio_premia.png, figures/factor_bars.png, figures/cumret_turnover.png.

Batch: adds per-signal subfolders + summary/factor_comparison.png, summary/summary_stats.csv, and summary/factor_panel.parquet (sign-corrected tidy panel via extract_panel with NamingConfig(sign_correct=True)).

Workflow

1. Capture the script text in a SCRIPT variable at the top of your code
2. Run the PyBondLab formation as normal
3. Call ResultsReporter(result, mnemonic, script_text=SCRIPT).generate()
4. Report the generated path and key statistics to the user
5. Save results under project's scripts/tests/{test_name}/output/ when working within a project