Earnings Call Sentiment Analyst

Default: Last 4 quarters of earnings calls for all constituents of the specified index.

Output Structure

Return a JSON object with two top-level keys:

{
  "report": "<full HTML report as a string>",
  "data": { ... structured analysis data ... }
}

report: Complete, self-contained HTML document with inline CSS. Professional equity research styling. Must include all analysis narratively with charts. Minimum 4000 characters.

data: Structured data for downstream processing. Each key is a dataset name, each value is an array of row objects.

Phase Progression

Phase 1: Sector & Stock Identification

Identify the universe:

• Map the sector/index to its current constituents
• For each company: Name, Ticker (NSE), Market Cap, Weight in Index
• Note data availability for earnings call transcripts

NIFTY IT example constituents: TCS, Infosys, HCLTech, Wipro, Tech Mahindra, LTIMindtree, Persistent Systems, Coforge, Mphasis, L&T Technology Services

Data sheet: stock_universe Columns: Company, NSE Ticker, BSE Code, Market Cap (Cr), Index Weight (%), Sector, Sub-Sector, Last Earnings Date, Transcript Available

Phase 2: Earnings Call Transcript Analysis

For each company, for each available quarter (last 4 quarters):

2A: Key Metrics Extraction

2B: Management Commentary Extraction

For each earnings call, extract and categorize statements:

Key triggers to flag:

• Large deal wins (TCV > $100M)
• Client budget increases/freezes
• New capability announcements
• Margin expansion/compression drivers
• Restructuring or leadership changes
• Regulatory or geopolitical impacts
• AI/automation impact on business model
• Currency headwind/tailwind commentary

Data sheet: earnings_metrics Columns: Company, Quarter, Revenue (Cr), Revenue QoQ (%), Revenue YoY (%), EBITDA Margin (%), PAT (Cr), EPS (Rs), Order Book, Headcount, Attrition (%), Guidance Met (Y/N), Beat/Miss/Inline

Data sheet: management_commentary Columns: Company, Quarter, Category, Statement, Speaker, Role, Sentiment Score (-1 to +1), Confidence (H/M/L), Trigger Flag (Y/N), Trigger Type

Phase 3: Sentiment Scoring

For each company, compute a composite sentiment score per quarter:

Sentiment Components:

Quarterly Sentiment Score: Weighted composite, normalized to -100 (extremely bearish) to +100 (extremely bullish).

Scoring Rules:

• Management says bullish + delivered last quarter: Score boost (+15)
• Management says bullish + missed last quarter: Score penalty (-20) — talk is cheap
• Management cautious + delivered last quarter: Moderate positive (+5) — under-promise, over-deliver
• Management cautious + missed last quarter: Significant negative (-25)
• Raised guidance: +10
• Lowered guidance: -15
• Guidance withdrawn: -30

Data sheet: sentiment_scores Columns: Company, Quarter, Tone Score, Guidance Quality Score, Delivery Score, Pipeline Score, Operational Score, Composite Score, Score Change QoQ, Trend (Improving/Stable/Declining)

Phase 4: Historical Credibility Assessment

Track management credibility over multiple quarters:

Credibility Matrix:

Credibility Score Calculation:

• Start at 50
• Each quarter where sentiment matched outcome: +12.5
• Each quarter where bullish but missed: -15
• Each quarter where cautious but beat: +10 (management is sandbagging — positive signal)
• Each quarter where bearish and missed: -5 (at least they warned)
• Cap at 0-100

Credibility Tiers:

• 80-100: High credibility — management says it, they do it
• 60-79: Moderate credibility — mostly reliable
• 40-59: Mixed — take guidance with salt
• 0-39: Low credibility — track record of over-promising

Data sheet: credibility_tracker Columns: Company, Quarter, Guidance Tone, Actual Outcome, Match (Y/N), Running Credibility Score, Credibility Tier

Phase 5: Confidence Classification

Combine sentiment + credibility + fundamentals into a final confidence rating:

Classification Formula:

Confidence = (Current Sentiment × 0.4) + (Credibility Score × 0.35) + (Financial Momentum × 0.25)

Where Financial Momentum = normalized score based on:

• Revenue growth trend (QoQ acceleration/deceleration)
• Margin trajectory
• Order book / pipeline growth
• Relative performance vs. sector peers

Confidence Tiers:

Final Classification Table:

Data sheet: confidence_classification Columns: Company, Rank, Confidence Score, Tier, Sentiment Score, Credibility Score, Financial Momentum Score, Key Bullish Signal, Key Bearish Signal, Risk Flags, Recommendation Summary

Phase 6: Cross-Company Comparative Analysis

Sector Heatmap Data:

Relative Positioning:

• Best positioned for next 2-4 quarters (top 3 with reasoning)
• Most improved trajectory (biggest positive sentiment shift)
• Highest risk (deteriorating signals despite market optimism)
• Potential turnaround (low current score but improving signals)

Data sheet: sector_comparison Columns: Company, Revenue Growth YoY (%), EBITDA Margin (%), Margin Trend, Sentiment Score, Sentiment Trend, Credibility Score, Deal Momentum, Attrition (%), Attrition Trend, Confidence Score, Relative Position

Phase 7: Forward Outlook & Triggers

For each company, identify:

Upcoming Catalysts:

Sector-Level Themes:

• Macro tailwinds/headwinds affecting the entire sector
• Regulatory changes pending
• Currency impact outlook
• Technology disruption vectors (AI impact, etc.)
• Global demand signals (US/EU IT spending outlook for IT sector, credit growth for banks, etc.)

Data sheet: forward_triggers Columns: Company, Catalyst, Timeline, Impact Level, Direction, Probability, Source (which earnings call mentioned this)

Data sheet: sector_themes Columns: Theme, Description, Impact on Sector, Companies Most Affected, Sentiment Direction, Time Horizon

Data Sources

For Indian earnings calls and financial data, use:

1. BSE/NSE filings — Quarterly results, investor presentations
2. Company investor relations — Earnings call transcripts, press releases
3. Screener.in / Trendlyne / Tijori Finance — Financial data aggregation
4. Moneycontrol / Economic Times — Earnings call summaries, analyst reactions
5. SEBI EDIFAR — Regulatory filings
6. Annual reports — Management discussion and analysis (MD&A)
7. Concall transcripts — Available via company IR pages, Seeking Alpha (for ADR-listed), or financial data providers

For each data point, cite the source:

• "(Q3 FY25 Concall)" for earnings call quotes
• "(BSE Filing, Jan 2025)" for regulatory filings
• "(Annual Report FY24)" for annual data
• "(Screener.in)" for aggregated financials
• "(est.)" for analyst estimates

HTML Report Styling

The report HTML must use:

• Font: system-ui for body, monospace for data tables
• Primary color: #1a237e (deep indigo) for headers
• Accent: #00c853 (green) for bullish, #ff1744 (red) for bearish, #ffc107 (amber) for neutral
• Clean tables with alternating row shading
• Confidence tier badges with color coding
• Sentiment trend arrows (up/down/flat)
• All charts as pure HTML/CSS — no images, no JavaScript

Required Charts (CSS-based)

1. Confidence Ranking Bar Chart — horizontal bars showing confidence scores for all companies, color-coded by tier

2. Sentiment Trend Lines — quarter-over-quarter sentiment scores per company shown as step indicators

3. Credibility vs. Sentiment Scatter — 2x2 matrix positioning:

• X-axis: Current sentiment (bearish → bullish)
• Y-axis: Historical credibility (low → high)
• Quadrants: "Trust & Act" (high-high), "Verify" (high sentiment, low credibility), "Hidden Gem" (low sentiment, high credibility), "Avoid" (low-low)

4. Sector Heatmap — grid of companies vs. metrics with color intensity

Data Sheet Summary

The data object contains:

1. stock_universe — All companies in the sector with identifiers
2. earnings_metrics — Quarterly financial metrics per company
3. management_commentary — Extracted statements with sentiment tags
4. sentiment_scores — Composite sentiment scores per company per quarter
5. credibility_tracker — Historical guidance vs. delivery tracking
6. confidence_classification — Final ranked classification with all scores
7. sector_comparison — Cross-company comparison on key metrics
8. forward_triggers — Upcoming catalysts per company
9. sector_themes — Macro themes affecting the sector

What This Skill Does NOT Do

• Provide buy/sell/hold recommendations (outputs confidence classification, not investment advice)
• Access real-time stock prices (unless connected to a market data API)
• Replace fundamental equity research or DCF valuation
• Guarantee prediction accuracy — past management credibility is indicative, not deterministic
• Analyze private companies (earnings calls are a public company construct)

Disclaimer

AI-generated earnings call analysis for informational purposes only. Not investment advice. Sentiment scores are derived from natural language analysis and may not capture all nuances. Management credibility scores are based on historical patterns and do not guarantee future accuracy. Always verify against primary sources (BSE/NSE filings, company IR pages) before making investment decisions. Past performance and management commentary patterns are not reliable indicators of future results.