IPOVision

IPOVision

Intelligent IPO Analytics Platform with ML-Based Prediction & Real-Time Sentiment Intelligence

---

Overview

IPOVision is a modular full-stack IPO analytics platform designed to combine machine learning, sector-aware intelligence, and real-time market sentiment monitoring into a unified financial analytics system.

The platform provides multiple independent analysis engines for evaluating IPO performance potential, market conviction, and sector-specific behavior through an interactive React dashboard backed by FastAPI microservices.

WebApp Link - https://ipo-project-website.vercel.app/

Built by FEC @ Indian Institute of Technology (IIT) Guwahati.

---

Key Features

• IPO Evaluation Engine: Predicts IPO return potential using financial and subscription metrics.
• Deep ML Analysis: Classification-based inference system with confidence probabilities.
• Sector-Specific Prediction: Dynamically routes requests to specialized sector-trained ML models.
• Real-Time Sentiment Streaming: Live WebSocket-powered sentiment analytics dashboard.
• Interactive Visualizations: Gauge meters, contribution graphs, conviction clusters, and velocity tracking.
• Modular Backend Architecture: Independent FastAPI services mounted into a unified backend.
• Scalable ML Infrastructure: Supports multiple serialized sklearn pipelines and future model expansion.

---

System Architecture

IPOVision follows a modular microservices-inspired architecture to ensure clean separation of concerns and scalable deployment.

1. Frontend Layer (React + Vite)

• Interactive multi-tab dashboard UI.
• Handles API communication and real-time WebSocket rendering.
• Provides visual analytics for all prediction engines.

2. Unified FastAPI Backend

Acts as the central orchestration layer.

Mounted Services

IPO Evaluator Service

• Structured IPO scoring pipeline.
• Financial and subscription analysis.
• Return projection visualization.

Deep Analysis Service

• Feature preprocessing and log transformation.
• ML classification pipeline.
• Confidence probability estimation.

Sector-Wise Prediction Service

• Sector-aware model routing.
• Specialized ML models for each market sector.
• Dynamic inference system.

Sentiment Analysis Service

• Background worker execution.
• WebSocket streaming.
• PostgreSQL LISTEN/NOTIFY integration.
• Live dashboard synchronization.

---

High-Level Data Flow

Frontend Dashboard (React/Vite)
            │
            ▼
Unified FastAPI Backend
            │
 ┌──────────┼───────────────┬───────────────┬───────────────┐
 ▼          ▼               ▼               ▼
IPO      Deep ML       Sector Models    Sentiment Engine
Engine   Analysis      Routing          (Realtime WS)

---

Sentiment Engine Architecture

The sentiment module uses asynchronous streaming architecture for real-time market intelligence.

User Request
     │
     ▼
FastAPI Endpoint (/sentiment/start)
     │
     ▼
Background Worker Thread
     │
     ▼
PostgreSQL LISTEN / NOTIFY
     │
     ▼
WebSocket Streaming Layer
     │
     ▼
Realtime React Dashboard

---

Tech Stack

Frontend

• React.js
• Vite
• Tailwind CSS
• Recharts
• WebSockets

Backend

• FastAPI
• Uvicorn
• AsyncIO
• PostgreSQL
• WebSockets
• Scikit-learn
• Joblib
• Pandas
• NumPy

Machine Learning

• Random Forest Regressor
• Linear Regression
• Feature Engineering Pipelines
• Sector-Specific Model Training

Deployment

• Vercel (Frontend)
• Railway / Render / Hugging Face Spaces (Backend)
• GitHub Actions Ready

---

Project Structure

IPO_PROJECT_WEBSITE/
│
├── backend/
│   │
│   ├── app/                     # IPO Evaluator Service
│   │   ├── services/
│   │   ├── main.py
│   │   ├── schemas.py
│   │   ├── model.pkl
│   │   └── constants.py
│   │
│   ├── deep/                    # Deep Analysis Service
│   │   ├── services/
│   │   ├── main.py
│   │   └── model.pkl
│   │
│   ├── sector/                  # Sector-Wise Prediction Service
│   │   ├── saved_models/
│   │   └── main.py
│   │
│   ├── sentiment/               # Real-Time Sentiment Engine
│   │   ├── worker.py
│   │   ├── main.py
│   │   └── app/
│   │
│   ├── main.py                  # Unified Backend Router
│   └── requirements.txt
│
├── frontend/
│   ├── src/
│   ├── public/
│   ├── package.json
│   └── vite.config.js
│
└── README.md

---

API Endpoints

IPO Evaluation

POST /ipo/predict

---

Deep Analysis

POST /deep/deep_analysis

---

Sector-Wise Prediction

POST /sector/predict

---

Sentiment Analysis

Start Worker

POST /sentiment/start

Stop Worker

POST /sentiment/stop

WebSocket Stream

/ws/v_t_stream

---

Installation Guide

Clone Repository

git clone https://github.com/Krishieboyy/ipo_project_website.git
cd ipo_project_website

---

Frontend Setup

cd frontend
npm install
npm run dev

Frontend runs on:

http://localhost:5173

---

Backend Setup

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

Backend runs on:

http://localhost:8000

---

Environment Variables

Frontend (.env)

VITE_API_BASE=http://localhost:8000
VITE_WS_BASE=ws://localhost:8000

---

Deployment

Frontend Deployment (Vercel)

npm run build

Deploy the frontend directory to Vercel.

---

Backend Deployment

Railway / Render

uvicorn main:app --host 0.0.0.0 --port $PORT

---

Future Improvements

• LLM-powered IPO summarization
• RAG-based financial document analysis
• Live NSE/BSE market data ingestion
• Explainable AI dashboards
• User authentication & portfolio tracking
• Historical IPO benchmarking
• Automated financial report parsing

---

Screenshots

IPO Evaluator

• Gauge-based return prediction
• Parameter contribution visualization
• Long-term and short-term analysis

Sentiment Dashboard

• Conviction clustering
• Sentiment velocity tracking
• AI confidence estimation
• Real-time streaming updates

Sector-Wise Prediction

• Dynamic sector selection
• Sector-specialized inference engine