Python SDK Guide

Import disclosure_alpha to extract sections, compute metrics, and score filings in notebooks or applications — without the CLI or HTTP layer.

Prerequisites: Installation; SEC EDGAR Setup for ticker helpers.

High-level helpers

Score local HTML:

from disclosure_alpha import score_filing_html

result = score_filing_html(open("filing.html").read(), "10-K")
print(result.scores.overall_disclosure_risk_score)
print(result.to_dict())

Score by ticker:

from disclosure_alpha import score_filing_ticker

result = score_filing_ticker("AAPL", 2025, form_type="10-K")
print(result.scores.overall_disclosure_risk_score)

Optional prior filing for diffs:

result = score_filing_html(html, "10-K", prior_html=prior_html)

Pipeline stages

Use lower-level functions when you need control over each step:

from disclosure_alpha import (
    extract_sections_from_html,
    compute_section_metrics,
    score_deterministic,
)

sections = extract_sections_from_html(html, form_type="10-K")
metrics = compute_section_metrics(sections, prior_sections=None)
scores = score_deterministic(metrics)

extract_sections_from_html returns a list of section objects with section_name, cleaned_text, word_count, and extraction_confidence.

Result fields

score_filing_* returns a ScoreResult with:

  • scores.overall_disclosure_risk_score — weighted headline (0–100)

  • scores.components — ten computed scores (nine headline-weighted plus specificity_quality_score; some may be None if sections are missing)

  • scores.confidence_score — derived from component coverage

  • scores.missing_components — list of components that could not be computed

See Deterministic Scoring Overview for component definitions and Section Taxonomy for section names.