MarketMind -- AI Crypto & Stock Market Intelligence¶
Status: Live demo Demo: market.robiriu-dev.my.id
Executive Summary¶
A market-intelligence SaaS for crypto and stocks. It pulls live market data, computes technical indicators, derives a Bullish / Bearish / Neutral signal, and has an AI layer that writes a balanced analysis for any asset and answers questions about it. The AI is grounded strictly on the computed indicators, so it never invents prices or statistics. Built as a polished dark trading terminal.
How It Works¶
Live data (CoinGecko crypto / pluggable stock feed) ─► 120-day price series
│
▼
Indicators (computed in code): SMA 20/50 · RSI(14) · volatility · 90d range
│ │
▼ ▼
Signal (SMA crossover + RSI) "data facts" context
Bullish / Bearish / Neutral │
├─► AI analysis (Gemini)
└─► Ask-the-market Q&A (grounded)Every number (price, RSI, moving averages, volatility, signal) is computed deterministically in TypeScript. Gemini only writes the narrative and answers questions, grounded on those facts, so the analysis is trustworthy.
Key Features¶
- Live crypto data - real prices and 120-day history for major coins (CoinGecko, keyless, cached); a Stocks tab with the same engine.
- Technical indicators - SMA 20 / 50, RSI(14) with overbought/oversold state, annualized volatility, and the 90-day high/low.
- Signal - a Bullish / Bearish / Neutral call from SMA crossover and RSI, with the reasons listed out.
- AI analysis - Gemini writes a concise, balanced read (trend, momentum, volatility, what the signal implies, and a risk caveat), using only the computed indicators.
- Ask the market - a chat that answers questions like "is it overbought?" or "how volatile is it?" from the real numbers.
- Resilient data layer - a deterministic fallback keeps the demo working even if a live feed is unavailable.
Technology Stack¶
| Layer | Technology |
|---|---|
| Frontend | Next.js 15 (App Router), TypeScript, Tailwind CSS (dark terminal UI) |
| Charts | Recharts (price area chart + sparklines) |
| Indicators | TypeScript (SMA, RSI, volatility, signal) -- deterministic, no LLM |
| LLM | Gemini 2.5 Flash via Vertex AI (analysis + grounded Q&A) |
| Data | CoinGecko (crypto, keyless); pluggable keyed feed for live stocks |
| Deployment | pm2 + nginx, Let's Encrypt SSL on a VPS |
To a market-ready SaaS¶
Add Stripe / PayPal subscriptions, user accounts with saved watchlists, a keyed live stock feed (Finnhub / Twelve Data), price alerts, and more assets. The indicator + signal + grounded-AI core stays the same.
Informational only, not financial advice.
Skills Demonstrated¶
- Integrating live financial data with a clean, resilient data layer
- Technical analysis in code (SMA, RSI, volatility, signal generation)
- Grounded LLM analysis and Q&A (no hallucinated numbers)
- Polished product UX (trading-terminal design, live indicators)
- Vertex AI integration and production deployment