What it does
The analyzer is a thematic research tool built on Bigdata.com. It:
- Searches news, filings, and transcripts for a configurable set of lenders and borrowers against predefined thematic queries (e.g. spread power, redemption pressure, maturity wall risk).
- Scores each entity by comparing how many on-topic, entity-relevant mentions fall into positive themes versus negative themes.
Configurable
- (Required re-run)
- Entities Lender and borrower names are configurable.
- Themes Search topics and their queries are configurable; You can define the risk and strength signals that matter to your investment process.
- Scoring The scoring formula can be replaced with your own weighting, composite, or rules-based model.
- Date range The search window is configurable; narrow to the last 30 days or widen to multi-year history.
Important
This is a technical showcase of Bigdata.com capabilities, not investment advice, a credit rating, or a trading signal. Scores are derived from mention counts after entity-in-text filtering and depend on the configured entities, topics, date range, and how content is indexed. Independent verification is recommended before any investment decision.
How to read scores
Counts are on-topic snippets where the entity name appears in the returned text, not raw unfiltered search volume.
Lender — Terms Power Score
terms_power_score = positive_count / (positive_count + negative_count + 1) × 100
Positive themes include spread power, fundraise resilience, NAV stability, and covenant tightening. Negative themes include redemption pressure, NAV markdown, covenant waivers, PIK stress, and fee compression.
Higher score = stronger lender positioning. A score near 100 means almost all entity-relevant mentions align with strength themes; a score near 0 means stress themes dominate.
Borrower — Stress Score
stress_score = 100 − terms_power_score
Positive (resilience) themes include revenue growth, refinancing success, margin expansion, and AI adoption. Negative (distress) themes include AI disruption risk, maturity wall, default/restructuring, and customer churn.
Higher score = greater distress signal. A score near 100 means distress themes heavily outweigh resilience mentions; a low score means the borrower is showing more positive coverage.
Who’s showing lender-strength narratives versus stress (redemptions, markdowns, waivers) in recent news and filings.
The radar plots intensity on stress topics only; the horizontal bars are Terms Power Score—higher means more strength-themed mentions relative to stress for that lender.
Stress radar
Each line is one stress topic (redemption pressure, NAV markdown, covenant waiver, PIK stress, fee compression). Value at each axis = mention count for that lender × topic. Same raw counts as the Signal Heatmap; only negative topics shown.
Terms Power Score by lender
Terms Power Score (0–100) = positive_mentions / (positive + negative + 1) × 100. Higher = stronger lender signal (spread power, fundraise resilience, NAV stability, covenant tightening vs redemption pressure, markdowns, waivers, PIK stress). Bars ranked by score descending.
Each cell counts snippets that matched the theme query and mention the lender in headline or body. Teal-headed columns are strength; coral-headed columns are stress—the same counts roll up into the score.
Signal Matrix
Each cell = number of returned documents where the lender’s name appears in the headline or content for that topic query (not the raw semantic-search list size). Rows = entities, columns = topics; these counts feed the Terms Power Score. Audit lists the same rows saved under .cache/scoring_audit/ when scores were computed.
Every theme is a fixed semantic search. Strength topics are listed first, then stress (green block above red). Each card shows its definition and scoring direction in parentheses.
Signal Topics & Queries
Supporting headlines and excerpts from the scoring run. Filter by entity or theme to verify what drove a cell.
Document Audit
End-to-end methodology: search, entity filter, polarity aggregation, and the Terms Power formula for this layer.
How Lender Scoring Works
The system combines hybrid semantic search, topic taxonomies, and polarity-based scoring to measure lender health.
✨ Why It Matters
Identifying which lenders hold strong negotiating positions versus those under redemption, markdown, or PIK stress is critical for understanding systemic private credit risk.
⚡ What It Does
Each lender is searched against strength topics (spread power, fundraise resilience, NAV stability, covenant tightening) and stress topics (redemption pressure, NAV markdown, covenant waivers, PIK stress, fee compression). Mention counts feed a Terms Power Score.
The Analysis Process
Topic Search
Each lender is searched against every signal topic using Bigdata semantic search. A sentiment filter matches each topic's polarity—positive topics retrieve predominantly positive-toned documents; stress topics retrieve predominantly negative-toned documents (same idea as API sentiment filters).
Entity Filtering
Results are filtered to only include documents where the entity name appears in headline or content.
Polarity Aggregation
Mention counts are split by polarity (positive / negative) and summed per entity.
Scoring & Ranking
Terms Power Score is computed and lenders are ranked descending. Radar chart shows negative-topic signal intensity.
🔢 Scoring Formula
terms_power_score = positive / (positive + negative + 1) × 100
High score = strong lender position. Low score = elevated stress signals relative to strength.
📅 Coverage
Search date filter: Jan 1, 2025 – Mar 30, 2026. Sources: news, filings, transcripts indexed by Bigdata.com.
Which PE-backed names skew toward resilience stories versus distress (maturity wall, default risk, churn, AI headwinds).
The radar highlights distress-topic volume by borrower; the stress bars combine positive and negative themes—higher means more distress signal.
Distress radar
Each line is one distress topic (AI disruption, maturity wall, default risk, customer churn). Value at each axis = mention count for that borrower × topic. Same raw counts as the Signal Heatmap; only negative topics shown.
Stress score by company
Stress score (0–100) = 100 − terms_power_score, where terms_power = positive / (positive + negative + 1) × 100. High score = more distress signal relative to resilience (revenue growth, refinancing). Ranked by stress descending.
Per-borrower, per-theme mention counts after the entity must appear in the returned text. Resilience columns first, then distress.
Signal Matrix
Each cell = number of returned documents where the borrower’s name appears in the headline or content for that topic—same rule as scoring. Positive columns (revenue growth, refinancing, etc.) and negative columns (AI disruption, maturity wall, default risk, customer churn) combine into the stress score as a ratio, not a raw sum. Audit reads the scoring snapshot (.cache/scoring_audit/), not raw search.
Queries used for borrowers: resilience themes on top, distress below. Each card shows its definition and scoring direction.
Signal Topics & Queries
Documents that passed the same filters used in scoring—use this tab to sanity-check a heatmap cell.
Document Audit
How borrower stress_score is built from positive versus negative theme counts and Bigdata retrieval settings.
How Borrower Distress Scoring Works
The system combines hybrid semantic search, risk factor taxonomies, and structured validation to transform unstructured data into actionable distress intelligence.
✨ Why It Matters
Understanding which PE-backed borrowers face AI disruption, maturity walls, default risk, or customer churn is critical for private credit portfolio monitoring.
⚡ What It Does
Each borrower is searched against resilience topics (revenue growth, refinancing success) and distress topics (AI disruption, maturity wall, default risk, customer churn). The ratio drives a Stress Score.
The Analysis Process
Topic Search
Each borrower is searched against every signal topic using Bigdata semantic search. A sentiment filter matches each topic's polarity—resilience topics favor positive-toned documents; distress topics favor negative-toned documents (same idea as API sentiment filters).
Entity Filtering
Results are filtered to only include documents that explicitly mention the borrower.
Polarity Aggregation
Mention counts are split by polarity and summed. Radar chart uses only negative topics.
Scoring & Ranking
Stress score is computed and borrowers are ranked descending. High score = most distressed.
🔢 Scoring Formula
stress_score = 100 − terms_power_score
High stress = more distress signal relative to resilience. Low stress = borrower showing strength signals.
📅 Coverage
Search date filter: Jan 1, 2025 – Mar 30, 2026. Sources: news, filings, transcripts indexed by Bigdata.com. Focus on PE-backed leveraged companies with private credit exposure.