Finmagine AI Advisor v2.0.0

SECTION 1

Introduction & Origin Story

If you've ever tried to get meaningful stock analysis from an AI chatbot, you know the frustration. You type "Is Amazon a good buy?" and you get a generic, hedged response that sounds like it was written by a compliance lawyer in 2021. The reason is simple: AI models are reasoning engines, but they're amnesiac. They only know what you tell them in that specific conversation, plus their training data — which is often months or years out of date.

The real bottleneck isn't intelligence. It's context. If you want institutional-grade analysis, you need to feed the AI current numbers: five years of balance sheets, cash flow trends, analyst consensus, valuation multiples, ownership data — all of it. And that leads to what we call copy-paste hell.

The Copy-Paste Problem

Picture the manual workflow: You open a tab for the stock price. Another for the income statement. A third for the balance sheet. Fourth for cash flow. Fifth for analyst ratings. You're jumping between tabs, highlighting rows, copying, pasting — and the formatting breaks every time. Columns don't align, tabs are wrong, the AI thinks the revenue number is actually a date because of some random spacing error. After 20 minutes of cleanup, you haven't even asked your first question.

Because this process is so painful, investors cut corners. They rely on gut feelings, headlines, or surface-level metrics. That's a dangerous way to invest.

Why a Separate Extension?

Finmagine already had a successful Chrome extension — the Finmagine Chart Builder — with 180+ users and a perfect 5.0 rating. The Chart Builder transforms financial tables into professional charts. So why build a completely separate extension for AI analysis?

The answer is risk isolation. The AI Advisor interacts with financial data platforms in a fundamentally different way than the Chart Builder. By keeping them as separate extensions, any issue with one product cannot affect the other. The Chart Builder's reputation and user base remain protected while the AI Advisor innovates independently.

The Finmagine Ecosystem

The AI Advisor doesn't exist in a vacuum. It's part of a broader Finmagine investment research ecosystem:

• Finmagine Chart Builder: Transforms financial tables into professional dual-axis charts (Screener.in, Google Finance, stockanalysis.com)
• Finmagine AI Advisor: Generates structured AI analysis prompts from real financial data (this extension)
• Finmagine Extension Analyst: A Custom GPT (powered by GPT-5.2) specifically trained with the Finmagine analysis methodology
• finmagine.com: The full platform with 71 company reports, 22+ interactive tools, rankings, and education

Together, these tools form a complete research workflow: visualize the data (Chart Builder), generate the analysis prompt (AI Advisor), and get expert-grade output (Custom GPT).

SECTION 2

How It Works — The Technical Magic

The Finmagine AI Advisor follows a three-step workflow that turns a 20-minute manual process into a 5-second click:

The User Flow

1. Visit any stock page on stockanalysis.com (e.g., stockanalysis.com/stocks/meta/)
2. The AI Advisor panel appears inline — embedded directly into the page, not a floating popup
3. Choose an analysis template — Comprehensive, Risk-Reward, or Quarterly Deep-Dive
4. A ~1,800-word structured prompt is generated with real financial data, computed metrics, and analysis instructions
5. Copy the prompt and paste it into ChatGPT, the Finmagine Extension Analyst GPT, or any AI assistant
6. Receive institutional-grade investment research

Data Extraction: Two Worlds

The extension supports two platforms with fundamentally different data architectures:

Aspect	Screener.in (India)	stockanalysis.com (US)
Extraction Method	Synchronous DOM scraping	Async 10-page parallel fetch
Network Requests	Zero — all data on page	10 fetches in staggered batches
Extraction Time	<100ms (instant)	~5–6 seconds
Rate Limiting	None	800ms batch gaps to avoid HTTP 429

The Analysis Engine

Before the prompt is even assembled, a local analysis engine (running entirely in your browser) computes several derived metrics:

• Health Score (0–100): A composite score reflecting overall financial health, calibrated to sector-specific thresholds
• CAGR Calculations: Revenue, profit, EPS, and free cash flow compound annual growth rates
• DuPont ROE Breakdown: Decomposing Return on Equity into margin, turnover, and leverage components
• Valuation Verdicts: Comparing current PE, P/B, and EV/EBITDA against historical medians
• Strengths & Concerns: Automatically identified based on data patterns

Prompt Assembly

The prompt builder merges raw financial data, computed analysis, and template-specific instructions into a structured ~1,800-word prompt. This isn't a simple "analyze this stock" request — it's a complete dossier with explicit scoring instructions, sector context, and data quality constraints that force the AI into a professional analytical persona.

SECTION 3

US/Global Stocks Deep Dive — stockanalysis.com

The Engineering Challenge

stockanalysis.com is built with Next.js — a modern single-page application framework. This creates two major engineering challenges that the extension must overcome:

Challenge 1: Fragmented Data. Unlike Screener.in (where all financial data lives on one page), stockanalysis.com splits data across 10 separate URLs. Each financial statement, each time period, each data category has its own page.

Challenge 2: SPA Navigation. When you click from one stock to another, the page doesn't reload. The website uses client-side routing to swap content without a full page refresh. A traditional Chrome extension (which runs once on page load) wouldn't even notice the change.

10-Page Parallel Fetch Architecture

The extension fetches data in two phases, with pages batched in pairs and 800ms gaps between batches to avoid rate limiting:

Phase 1 — Core Annual Data (5 pages):

Page	Data Extracted
Annual Income Statement	Revenue, Net Income, EPS, Margins, EBITDA
Annual Balance Sheet	Assets, Liabilities, Equity, Debt
Annual Cash Flow	Operating CF, CapEx, Free Cash Flow
Annual Ratios	PE, P/B, EV/EBITDA, ROE, ROA
Company Profile	Sector, Industry (GICS), Description

Phase 2 — Quarterly & Enrichment Data (5 pages):

Page	Data Extracted
Quarterly Income Statement	Quarterly Revenue, Net Income, EPS
Quarterly Balance Sheet	Quarterly Assets, Debt
Quarterly Cash Flow	Quarterly OCF, FCF
Statistics	50+ metrics: Z-Score, F-Score, ROIC, WACC, analyst data, ownership
Price History	Daily OHLCV data

Instant UI Shell Pattern

Six seconds of blank screen is an eternity in UX terms. The extension solves this with an instant UI shell: the moment you load a stock page, the panel renders immediately with the three template cards visible. While you're reading the template descriptions and deciding which analysis you want, the 10 pages are fetching in the background. By the time you click a card, the data has usually arrived.

Advanced US-Exclusive Metrics

Altman Z-Score — Bankruptcy Risk Assessment

Developed by Professor Edward Altman in 1968, the Z-Score combines five financial ratios to predict bankruptcy probability within two years. The extension extracts the pre-computed value from the Statistics page.

Zone	Z-Score	Meaning
Safe	> 2.99	Low bankruptcy risk
Grey	1.81 – 2.99	Uncertain territory
Distress	< 1.81	High bankruptcy risk

Piotroski F-Score — Financial Strength (0–9)

A 9-point binary scoring system evaluating profitability (4 criteria), leverage (3 criteria), and efficiency (2 criteria). Each criterion the company passes earns one point. Scores of 7–9 indicate strong financial health; 0–3 indicate weakness.

ROIC-WACC Spread — The North Star Metric

The difference between Return on Invested Capital (what the company earns) and Weighted Average Cost of Capital (what it costs to fund investments). This is the purest measure of value creation:

Spread	Meaning	Signal
> +15%	Exceptional value creation	Strong moat — Core Compounder
+5% to +15%	Solid value creation	Healthy capital allocation
0% to +5%	Marginal value creation	Adequate but not exceptional
Negative	Value destruction	Investigate if temporary or structural

Analyst Consensus & Price Targets

The extension pulls Buy/Hold/Sell distribution, mean/high/low price targets, and forward estimates. The AI uses a "confirm vs. contradict" logic: when strong fundamentals align with a Buy consensus, conviction increases. When they diverge, the AI flags the discrepancy and investigates.

Institutional & Insider Ownership + Short Interest

Three ownership signals provide context on who owns the stock and who's betting against it. High insider ownership signals management alignment; high short interest signals bearish conviction that warrants investigation.

145 GICS Industry Classifications

US stocks use the Global Industry Classification Standard (GICS) with 11 sectors, 24 industry groups, and 145 distinct industries. The extension contains a mapping layer that translates all 145 GICS industries to its 18 core analysis profiles. This ensures sector-appropriate thresholds are applied: a bank isn't penalized for high leverage, and a tech company isn't penalized for a high PE ratio.

The Three US Templates

Comprehensive Analysis — The 360-Degree View

Best for first-time analysis or annual portfolio reviews. Produces a complete investment thesis covering financial health, growth, competitive positioning, management quality, and valuation. Use this when you need the full picture.

Risk-Reward Analysis — The Decision Maker

Best for buy/sell decisions and position sizing. Explicitly models bull/base/bear scenarios with probability weights and calculates asymmetry ratios. Use this when you're hovering over the buy or sell button.

Quarterly Deep-Dive — The Monitor

Best for post-earnings analysis and trend monitoring. Focuses on sequential quarter-over-quarter changes rather than multi-year averages. Use this when you own a stock and want to know if the thesis is still intact.

SECTION 4

India vs US — Platform Comparison

The AI Advisor works on both Screener.in (Indian stocks) and stockanalysis.com (US/Global stocks), but the analysis depth differs significantly. It's a trade-off between qualitative depth (India) and quantitative breadth (US).

Aspect	India (Screener.in)	US (stockanalysis.com)
Templates	5 (Comprehensive, Risk-Reward, Management Quality, Quarterly, Deep Research)	3 (Comprehensive, Risk-Reward, Quarterly)
Extraction	Synchronous DOM scraping (instant)	Async 10-page parallel fetch (~5–6s)
Financial History	10+ years	5 years (free tier)
Quarterly Data	P&L only	P&L + Balance Sheet + Cash Flow
Financial Scores	Not available	Z-Score, F-Score, ROIC-WACC
Analyst Data	Not available	Consensus, price targets, forecasts
Ownership	Promoter/FII/DII	Institutional + Insider + Short Interest
Conference Calls	12 quarters (BSE PDF links)	Not available
Annual Reports	All available (BSE links)	Not available
Credit Ratings	All available	Not available
Sector Profiles	18 sectors	145 GICS industries → 18 profiles

Why India Has 5 Templates but US Has 3

The Management Quality template analyzes concall transcripts, management guidance vs. actual delivery, and governance signals. The Deep Research template instructs the AI to browse BSE PDF links (concalls, presentations, annual reports) for a 15–30 minute forensic analysis. Both require document URLs that stockanalysis.com doesn't provide.

SECTION 5

The Finmagine Framework — Scoring Methodology

Every analysis the AI Advisor generates follows a standardized scoring methodology. This isn't a black box — the framework is explicit, weighted, and consistent across companies, enabling apples-to-oranges comparison between a tech giant and an energy major.

The Five-Parameter Weighted Score

Parameter	Weight	What the AI Evaluates
Financial Health	25%	Z-Score, F-Score, D/E ratio, current ratio, interest coverage, cash flow quality
Growth Prospects	25%	Revenue CAGR, profit CAGR, FCF CAGR, margin trajectory, reinvestment rate
Competitive Position	20%	ROIC-WACC spread, margin sustainability, market share indicators
Management Quality	15%	Capital allocation (ROIC trend), insider ownership, share dilution/buybacks
Valuation	15%	PE vs history, P/FCF, EV/EBITDA, PEG, price target vs current

Composite Score = (Financial Health × 0.25) + (Growth × 0.25) + (Competitive Position × 0.20) + (Management × 0.15) + (Valuation × 0.15)

Score Interpretation

Composite	Rating	Typical Action
8.5 – 10.0	Exceptional	Strong conviction buy candidate
7.0 – 8.4	Strong	Solid investment candidate
5.5 – 6.9	Average	Requires catalyst or deep value thesis
4.0 – 5.4	Below Average	Significant concerns
< 4.0	Weak	Avoid unless contrarian thesis is very strong

Investment Classifications

Based on the Five-Parameter profile, the AI assigns one of five classifications:

• Core Compounder: High across all parameters. Dominant market position, strong ROIC-WACC, consistent growth, reasonable valuation. (Example: AMZN 8.2/10)
• Growth Compounder: Growth Prospects dominant. Accelerating revenue, expanding margins, may trade at premium. (Example: UBER 7.6/10)
• Cyclical Compounder: Strong Financial Health + Competitive Position; Growth varies with industry cycles. (Example: LRCX)
• Capital Preservation Compounder: Financial Health + Valuation strong. Fortress balance sheet, modest growth, downside protection. (Example: BRK.A 8.0/10)
• Value-Growth Hybrid: Attractive valuation + re-accelerating growth. Transitional phase. (Example: PDD 8.8/10)

Sector-Aware Thresholds

A 15% ROE means very different things in different industries. For an IT company, it's mediocre. For a bank, it's excellent. For a utility, it's exceptional. The framework applies sector-specific thresholds so that a banking stock isn't unfairly penalized for high leverage, and a tech stock isn't unfairly penalized for a high PE ratio.

SECTION 6

Real-World Case Studies

Let's walk through three complete examples showing the full journey from stock page to GPT output. Each company represents a fundamentally different business model, demonstrating how the same framework adapts to wildly different situations.

Case Study 1: Amazon (AMZN) — Comprehensive Analysis

Health Score: 93/100 (Excellent)

Amazon was classified as a Core Compounder with a weighted score of 8.2/10. Here's how the five parameters scored:

Parameter	Score	Key Driver
Financial Health	9/10	Z-Score 5.06 (Safe), massive OCF ($139.5B), rapid deleveraging
Growth Prospects	8/10	Revenue CAGR 8.8% (5Y), profit CAGR 18.4%, EPS forecast 19.3%
Competitive Position	9/10	ROIC-WACC spread +3.2%, wide moat (network effects + switching costs)
Management Quality	8/10	9.01% insider ownership, disciplined restructuring post-2022
Valuation	7/10	PE 27.69, Forward PE 25.78, PEG 1.49 — not cheap but reasonable

The AI correctly identified that while ROIC-WACC spread of +3.2% is lower than a pure software company, it's highly acceptable for Amazon's capital-intensive infrastructure model. Every dollar Amazon invests in AWS data centers and fulfillment centers generates returns above its cost of capital.

Case Study 2: Berkshire Hathaway (BRK.A) — Risk-Reward Analysis

Health Score: 100/100 (Exceptional)

Berkshire was classified as a Capital Preservation Compounder with a weighted score of 8.0/10.

Scenario	Probability	Drivers	12–18M Return
Bull	25%	Investment gains rebound, insurance pricing strong, buybacks accelerate	+15% to +25%
Base	50%	Stable underwriting, moderate investment returns	0% to +8%
Bear	25%	Market drawdown, insurance losses, earnings compression	-15% to -25%

The most fascinating aspect of the Berkshire analysis was the Z-Score handling. Berkshire's Z-Score was 2.88 — technically in the "Grey Zone" that borders financial distress. A naive system would have flagged this as a major red flag for a company sitting on $334 billion in cash.

But the extension's sector-aware engine recognized Berkshire as an insurance/financial holding company. Insurance companies carry massive float liabilities by design — it's their business model. The AI explicitly stated: "The Altman Z-Score Grey Zone classification reflects large insurance liabilities and mark-to-market accounting volatility. Given the net cash position of +$203.4 billion, distress risk appears low despite the Grey classification."

ROIC-WACC Spread: +9.03% — Exceptional capital efficiency. Even with Berkshire's conservative posture, every dollar invested generates returns far above the cost of capital. The AI used this as the primary input for the Capital Preservation classification.

Case Study 3: PDD Holdings (PDD) — Quarterly Deep-Dive

Health Score: 100/100 (Exceptional)

PDD was classified as a Value-Growth Hybrid with a weighted score of 8.8/10 — the highest of our three case studies. But the story underneath was far more nuanced than the headline score suggests.

The Quarterly Deep-Dive template captured a dramatic growth deceleration trajectory:

Quarter	YoY Revenue Growth	Operating Margin
Q1 2024	130.66%	25.22%
Q2 2024	85.65%	33.55%
Q3 2024	44.33%	24.36%
Q4 2024	24.45%	24.83%
Q1 2025	10.21%	16.81%
Q2 2025	7.13%	24.80%
Q3 2025	8.98%	23.11%

From triple-digit growth to single digits in 18 months. Margin compression from 33.55% peak to 23.11%. This is the classic trap for growth stock investors: the PE ratio (10.48x) looks incredibly cheap, but only if growth re-accelerates. If growth is permanently stuck at 8–9%, then 10.48x is just fair value — possibly expensive given geopolitical risks.

The AI correctly identified this tension: "PDD's quarterly momentum shows growth normalization but sustained profitability strength. If growth stabilizes near 8–12% while maintaining >20% operating margins, current valuation appears attractive. The stock remains a high-quality value-growth hybrid, but hypergrowth days appear behind it."

Comparing the Three Case Studies

Company	Template	Classification	Composite	Key Metric
AMZN	Comprehensive	Core Compounder	8.2/10	ROIC-WACC +3.2% (infra-adjusted)
BRK.A	Risk-Reward	Capital Preservation	8.0/10	Net cash +$203B, Z-Score contextualized
PDD	Quarterly	Value-Growth Hybrid	8.8/10	Growth deceleration 131% → 9%

Three very different companies, three different templates, three different investment stories — all analyzed through the same consistent Five-Parameter Framework. This is the power of standardized methodology: you can compare a tech infrastructure giant, a financial conglomerate, and a Chinese e-commerce company on the same analytical plane.

SECTION 7

Privacy, Security & Trust

When a browser extension reads financial data, privacy concerns are legitimate. Here's why the Finmagine AI Advisor is architecturally different from most extensions:

100% Client-Side Processing

There is no Finmagine server collecting your data. The extension is a piece of JavaScript that runs entirely inside your browser on your own machine. The data flow is:

stockanalysis.com → Your Browser (extraction + analysis) → Your Clipboard → Your AI Chat

Finmagine never touches it. No API keys, no tracking, no middleman.

Zero Data Collection

• No analytics: No Google Analytics, no Mixpanel, no event tracking
• No telemetry: The extension doesn't phone home
• No account: No login, no registration, no email collection
• No external servers: Zero network requests to Finmagine infrastructure

Chrome Manifest V3 Security Model

The extension uses Chrome's latest Manifest V3 security model with minimal permissions. Host permissions are limited to the two supported sites (screener.in and stockanalysis.com). No access to browsing history, bookmarks, or other tabs.

SECTION 8

Tips, Tricks & Power User Guide

Edit Prompts Before Copying

Click the Edit button to toggle the prompt textarea between readonly and editable mode. You can add specific questions, remove sections you don't need, or adjust the analysis focus before copying. Your edits persist until you regenerate or switch templates.

Use Any AI Assistant

While the prompts are optimized for the Finmagine Extension Analyst GPT, they work with any AI model. The structured format and explicit instructions ensure high-quality output regardless of which assistant you use. Try ChatGPT, Gemini, or any other conversational AI.

Combine with Finmagine Chart Builder

For the ultimate research workflow: use the Chart Builder to visualize the financial trends, then use the AI Advisor to generate the analysis prompt. Visual pattern recognition (charts) combined with structured analytical reasoning (AI) gives you both the intuitive and the rigorous perspectives.

Template Selection Guide

Your Question	Best Template	Why
"Should I buy this stock?"	Risk-Reward	Explicitly models upside/downside scenarios
"Tell me everything about this company"	Comprehensive	Full 360-degree view
"How was the latest quarter?"	Quarterly	Sequential trends and recent momentum
"Annual review of my holdings"	Comprehensive	Complete reassessment
"Is this turnaround real?"	Risk-Reward	Quantifies asymmetry between recovery and decline

Understanding AI Confidence Levels

The AI's confidence is highest when multiple signals align: strong health score + positive ROIC-WACC spread + Buy consensus + high F-Score. When signals diverge (e.g., strong fundamentals but analyst downgrades), the AI will explicitly flag the disagreement. Pay special attention to these divergence moments — they're often where the best investment insights hide.

The Data Quality Constraints Block

Every prompt includes a constraints section that instructs the AI to: flag any data gaps or inconsistencies, avoid hallucinating numbers not present in the data, apply sector-specific context, and distinguish between reported facts and forward-looking estimates. This acts as a guardrail against the most common AI failure modes.

SECTION 9

The Future — What's Next

Planned for Future Versions

• Google Finance Support: Bringing the AI Advisor to Google's stock pages for broader accessibility
• Additional AI Platforms: Direct integration with Perplexity, Gemini, and other AI assistants beyond ChatGPT
• More Prompt Templates: ESG analysis, competitive positioning, detailed DuPont decomposition, and dividend sustainability templates
• Peer Comparison Prompts: Generate prompts that compare 2–3 companies side-by-side for relative valuation
• Earnings Preview Mode: Pre-earnings analysis with consensus expectations and historical beat/miss patterns
• Chart Builder Integration: Deep linking between the AI Advisor and Chart Builder extensions for a unified research workflow
• Additional Exchanges: Support for LSE, TSX, ASX, and other international stock markets

The Finmagine AI Advisor is evolving from a single-purpose prompt generator into a comprehensive research companion that covers the full investment analysis lifecycle. Every update is driven by the same core principle: real data in, expert analysis out.

Your Turn — Start Analyzing

The barrier to performing institutional-quality stock analysis has fundamentally collapsed. What used to require a Bloomberg terminal, a team of analysts, and 30 minutes of manual data gathering now happens with a single click in your browser.

What You've Learned:

• How the AI Advisor extracts data from 10 pages in staggered batches
• What Altman Z-Score, Piotroski F-Score, and ROIC-WACC spread reveal about a company
• How 145 GICS industries map to sector-aware analysis profiles
• The Five-Parameter Scoring methodology and investment classifications
• Real case studies: AMZN (Core Compounder), BRK.A (Capital Preservation), PDD (Value-Growth Hybrid)
• Why privacy-first architecture matters for financial research

Install the extension: Search for "Finmagine AI Advisor" on the Chrome Web Store or visit finmagine.com for more tools and resources.