Building a Stock Analysis Engine: From Zero to Buy/Sell Signals in 2 Weeks

#javascript #webdev #productivity #beginners

Building a Stock Analysis Engine: From Zero to Buy/Sell Signals in 2 Weeks

The Challenge

MoodFlow already had 5 AI content tools. But our users in Singapore, Malaysia, and Indonesia kept asking: "Can you add stock analysis?"

The requirements were brutal:

Instant signal — not charts, not reports. A verdict.
Multi-market — SGX, Bursa Malaysia, IDX
Deep analysis — 18+ technical indicators + fundamentals + capital flow
Free tier — daily picks at zero cost

The Tech Stack

Layer	Choice	Why
Data	yfinance	Free, covers SG/MY/ID markets
Scoring	Custom Python engine	PE + ROE + revenue growth + OBV + MFI + volume ratio
AI Analysis	DeepSeek API	Deep analysis when rule engine shows ambiguity
API	FastAPI	Async, lightweight, JSON responses
Deployment	Singapore VPS	Low latency to SEA users

The Scoring Engine

Total Score = Technical(40%) + Fundamental(35%) + Capital Flow(25%)

Technical:  buy/sell signals × 5 → 0-100
Fundamental: PE(50pts) + ROE(30pts) + Revenue Growth(20pts) → 0-100  
Capital:     OBV(40pts) + MFI(35pts) + Volume Ratio(25pts) → 0-100

Each indicator is weighted and cross-checked. When the rule engine and AI disagree — that's a "deep divergence" signal, which actually tells you something interesting is happening.