StockSense is a self-hosted trading intelligence platform that combines live market data, ML price forecasts, news sentiment analysis, and end-to-end portfolio management — built for retail investors who want a private, extensible alternative to premium terminals.
Six core feature areas covering the full investment workflow, from data ingestion to tax reporting.
Prophet + Gradient Boosting ensemble trained on 32+ features. Confidence-tiered predictions with inverse-MAPE blending. Auto-retrains every 24 hours per symbol.
Polls 6 RSS feeds every 15 minutes. VADER sentiment scoring with finance-specific lexicon. 7-day and 30-day rolling averages feed directly into the prediction model.
FIFO cost basis across all holdings. Realized and unrealized P&L breakdown. AI-powered advisor scores each position for buy/hold/sell signals.
Real-time quotes and OHLCV history for NSE, BSE, and NASDAQ via yfinance. 15-minute cache with background polling during market hours (09:15–15:30 IST).
Define signal conditions using RSI, MACD, or Bollinger Bands. Run backtests against historical data. Review equity curve, Sharpe ratio, and full trade log.
Capital gains classified as short-term or long-term under Indian tax rules. Automatic ₹1 lakh LTCG exemption. Tax-loss harvesting suggestions with savings calculated.
Three stages, fully automated. You make the decision; StockSense handles the analysis.
Autonomous data-fetcher agent pulls live OHLCV quotes and news from 6 RSS feeds every 15 minutes during market hours. Everything is normalized and stored locally.
ContinuousA blended Prophet + GBM model assembles 32+ features per symbol — candlestick patterns, technical indicators, sentiment, macro data — and generates 30-day price forecasts.
Every 24 hoursReview predictions with confidence tiers, screener results, portfolio P&L, strategy backtest outcomes, and AI-generated signals. The advisor explains its reasoning.
On demandRequires Python 3.11+, Node 20+, and ta-lib (brew install ta-lib on macOS).
Pull the source code and enter the project directory.
Copy the example env file. The only required variable to get started is JWT_SECRET_KEY.
The start.sh script boots both backend and frontend. Default login is admin / changeme.
# 1. Clone git clone https://github.com/rupakc/stocksense.git cd stocksense # 2. Configure cp .env.example .env # Set JWT_SECRET_KEY (required) # 3. Run ./start.sh # Backend API → http://localhost:8000 # Frontend → http://localhost:5173 # API docs → http://localhost:8000/docs # Login → admin / changeme
Full installation guide including ta-lib setup and GCP deployment in the project wiki.
Clone the repo, point it at your watchlist, and let the models do the heavy lifting.
Everything you need to understand, extend, and deploy StockSense.