VaLuLens — Stock Intrinsic Value Finder

VaLuLens is a web-based fundamental analysis platform that helps investors estimate the intrinsic value of any stock using multiple valuation methods, health scoring, and machine learning growth predictions.

Live: funda.farelhanafi.my.id

---

Features

• Multi-Method Valuation — Combines Graham Number, Discounted Cash Flow (DCF), and P/E Band multiple into a weighted fair value estimate.
• Global Stock Support — Fetches real-time data from Yahoo Finance for 10+ exchanges: IDX, NYSE/NASDAQ, SGX, Bursa Malaysia, TSE, HKEX, SSE/SZSE, LSE, ASX, KOSPI/KOSDAQ.
• Health Checklist — Instant evaluation based on ROE, D/E ratio, P/B, P/E, EPS growth trend, and dividend yield.
• ML Growth Prediction — TensorFlow.js LSTM model trained on 25+ Indonesian stocks. Uses multi-feature input (EPS, ROE, D/E, dividend yield) to predict next-year EPS growth.
• Buy/Sell Price Zones — Calculates ideal buy price, buy range, and sell target based on intrinsic value and margin of safety.
• Stock Condition Summary — Narrative verdict (Strong Buy / Buy / Hold / Sell / Avoid) with positives, risks, and sector-specific notes.
• Financial Health Score — 0-100 score combining ROE efficiency, debt level, and growth trajectory.
• Export & Share — Save analysis as PNG image or share via URL with embedded parameters.
• Side-by-Side Comparison — Compare two stocks across all metrics.

---

Methodology

Valuation Models

1. Graham Number — sqrt(22.5 x EPS x BVPS). Conservative fair value for companies with positive earnings and book value.

2. DCF (Discounted Cash Flow) — Projects EPS forward 5 years using historical CAGR (or ROE-based estimate as fallback), then discounts back at 10% (12% for banks). Terminal value uses Gordon Growth Model.

3. P/E Band — EPS x sector-adjusted P/E. Uses 15x for most sectors, 12x for banks.

4. Weighted Fair Value — Combines all three: 30% Graham + 40% DCF + 30% P/E Band.

ML Model

• Architecture: 2-layer LSTM (64 + 32 units) with dropout, dense output.
• Features: 4 per timestep — EPS, ROE, D/E ratio, dividend yield.
• Training: 80 epochs on 25+ Indonesian stocks from Yahoo Finance (10-year annual data).
• Prediction: Takes last 3 years of multi-feature data, outputs next-year EPS growth rate.
• Persistence: Saved to IndexedDB, survives page reloads.

Bank-Aware Logic

• D/E ratio is excluded from scoring for banks (deposits are liabilities, not debt).
• Bank discount rate is 12% (vs 10% for non-banks).
• P/E fair value threshold is 12x (vs 15x for non-banks).

---

Tech Stack

Layer	Technology
Framework	Vite 8 + React 18 + TypeScript
Styling	Tailwind CSS v4 + shadcn/ui (New York style)
ML	TensorFlow.js (LSTM, runs in-browser)
Data	Yahoo Finance API via yahoo-finance2
Routing	wouter
Forms	react-hook-form + zod
Charts	Recharts
Deployment	Vercel (serverless functions)
Fonts	Instrument Serif, DM Sans, IBM Plex Mono

---

Getting Started

# Install dependencies
pnpm install

# Run dev server
pnpm dev

# Build for production
pnpm build

The dev server runs at http://localhost:5173. API endpoints are in the api/ directory and run as Vercel serverless functions.

---

Project Structure

VaLuLens/
  api/
    stock.mjs          # Single stock data endpoint (Yahoo Finance)
    stocks-batch.mjs   # Batch endpoint for ML training data (25+ stocks)
  src/
    lib/
      calculations.ts  # Valuation math, health scoring, price zones
      ml-model.ts      # TensorFlow.js LSTM model (train/predict/save/load)
      exchanges.ts     # Global exchange definitions (suffix, examples)
      currency.ts      # Currency formatting by ticker suffix
    pages/
      Home.tsx         # Landing page with methodology overview
      Analyze.tsx      # Main analysis page (fetch, train, predict, calculate)
      Compare.tsx      # Side-by-side stock comparison
      About.tsx        # Methodology details and disclaimers
    components/
      layout/          # Navbar, Footer
      analyzer/        # EpsChart component
  index.html           # Entry point with SEO meta tags

---

Usage

1. Select an Exchange from the dropdown (default: Indonesia / IDX).
2. Enter a Ticker Symbol (e.g., BBCA, AAPL, 0700).
3. Click the Search button to fetch real-time data from Yahoo Finance.
4. The analysis calculates automatically — view intrinsic value, health checklist, buy/sell zones, and condition summary.
5. Optionally Train the ML model for EPS growth predictions.
6. Export as image or Share via link.

---

Disclaimer

VaLuLens is an educational and decision-support tool, not professional financial advice. Stock investing carries significant risk. Always do your own due diligence before making investment decisions.

---

Developed by Farel Hanafi (c) 2026