Advanced AI-Driven Dashboard

Beacon

An AI-driven portfolio & market intelligence dashboard that synthesizes technical indicators, live news sentiment, and LLM reasoning into opinionated trading recommendations.

View DemoView on GitHubExplore features
Beacon Dashboard

What's inside

Core Features

Market DashboardLive Data

Market Dashboard

Portfolio & P/L

Real-time P/L tracking against Postgres buying prices. Visualizes portfolio value, 24h market volume, and a live Fear & Greed Index gauge.

Global News StreamWebSocket

Global News Stream

FinBERT Sentiment

Live WebSocket news with automatic macro sentiment analysis (Positive vs Neutral vs Negative) powered by FinBERT NLP model.

AI RecommendationsAI-Powered

AI Recommendations

RAG Pipeline Output

MACD, Bollinger Bands, Moving Averages, and Volatility — with the AI's exact reasoning bullets and an action conviction score.

Liquidation TrackerRisk

Liquidation Tracker

Derivatives & Futures

Funding Rates, Open Interest, Long/Short Ratios. Proprietary Liquidation Heat score warns of impending long or short squeezes.

India Crypto HubLocal

India Crypto Hub

Arbitrage & Tax

Measures the 'India Premium' gap between global USD and local INR prices. Includes a tax-aware sell preview for Indian users.

DCA SimulatorSimulator

DCA Simulator

Backtesting Engine

Backtest Dollar Cost Averaging strategies with exact buy points on daily/weekly/monthly intervals. Renders a precise Recharts AreaChart.

System Architecture

Component Diagram

Frontend
Backend Server
Databases
External Data Sources
Cognitive Layer
REST & WebSocket
Calculate RSI/MACD
Store/Retrieve JSON
Fetch Prices/OHLCV
Fetch Breaking News
Analyze Sentiment
Generate Recommendations
Next.js AppRouter UI
Redis / MemoryCache
Express Node.jsServer
DeterministicMath Engine
PostgreSQL +Prisma
Binance /CoinGecko /DexScreener
RSS / GlobalNews APIs
HuggingFaceFinBERT - NLP
Qwenqwen3.8-27b - LLM

How it thinks

AI Cognitive Engine

Beacon's recommendation engine uses a rigorous Retrieval-Augmented Generation (RAG) pipeline tailored specifically for quantitative finance — no hallucination, only grounded math.

01

Data Ingestion

Queries external APIs for exact 30-day OHLCV data, global market cap, and recent news articles.

02

Deterministic Math

Computes exact technical indicators (RSI, MACD, Bollinger Bands) server-side to feed factual math to the LLM.

03

NLP Sentiment

Feeds news titles/descriptions into FinBERT to get a weighted sentiment score from -1 to 1.

04

LLM Synthesis

Forces the LLM to output a strict JSON schema with price targets, actions, and exact reasoning.

INGESTION
MATH AND NLP
LLM SYNTHESIS
PERSIST & DISPLAY

Price Data

1A
  • 30D OHLCV
  • Live Prices

News Data

1B
  • News APIs
  • RSS Feeds

Math Engine

2
[ COMPUTED, NOT GENERATED ]
  • RSI / MACD
  • Bollinger Bands

NLP Sentiment

3
  • FinBERT Model
  • Score -1 to 1

Context Build

4
  • Merge Signals
  • Format Data

Qwen 3.8

5
  • qwen3.8-27b
  • Market Reasoning

Strict JSON

6
  • BUY / HOLD / SELL
  • Confidence %
BUYHOLDSELL
*Not financial advice

Storage

7
  • PostgreSQL
  • JSONB Schema

Dashboard

8
  • UI Update
  • Simulators
PRICE_LANE
NEWS_LANE

Data layer

JSONB Advantage

Beacon uses PostgreSQL's JSON/JSONB column type via Prisma to handle unstructured LLM outputs without rigid schema migrations.

When the AI prompt is updated to return new indicators, the database adapts automatically — everything is preserved exactly as the model generates it.

schema.prisma
model PortfolioAnalysis {
  id          String    @id @default(uuid())
  portfolioId String
  portfolio   Portfolio @relation(
    fields: [portfolioId],
    references: [id],
    onDelete: Cascade
  )

  data        Json     // Full AI output (JSONB)
  createdAt   DateTime @default(now())
}

Self-host

Getting Started

Prerequisites: Node.js v18+, PostgreSQL, Qwen API Key & HuggingFace token.

1

Clone the repository

git clone https://github.com/punyajain1/Beacon.git
2

Setup Backend

cd backend
npm install
# Setup .env with DATABASE_URL, QWEN_API_KEY, etc.
npx prisma db push
npm run dev
3

Setup Frontend

cd frontend
npm install
# Setup .env.local if needed
npm run dev
4

View Dashboard

Navigate to http://localhost:3000
to view the Beacon Dashboard.
View full source on GitHub