GPST — Global Predictive Stock Terminal

AI-powered stock intelligence terminal built entirely on free and open data sources.

Live Demo · GitHub · Buy Me a Coffee

---

Disclaimer: GPST does not provide financial advice. All predictions are generated by automated models for educational and informational purposes only. Past performance does not guarantee future results. Always consult a licensed financial advisor before making investment decisions.

---

What it does

GPST tracks 90+ global tickers (US equities, international stocks, ETFs, sector and commodity proxies) and generates daily directional predictions using a combination of technical analysis, news sentiment, and macroeconomic signals.

Each ticker receives a prediction (UP / DOWN / NEUTRAL), a confidence score, and a signal explanation. Predictions are evaluated against the next market close and accuracy is tracked over time.

Data sources

All data comes from free and open APIs. GPST works even without API keys using Yahoo Finance and GDELT.

Source	Provides	Key required
Yahoo Finance	Price data, OHLCV	No
GDELT	Global news sentiment	No
FRED	VIX, yield curve, Fed Funds Rate	Optional
Alpha Vantage	News sentiment, quotes	Optional (~25/day)
NewsAPI	Headlines	Optional (100/day)

---

Prediction engine

GPST supports two prediction architectures.

v1 — Rules-based model

A deterministic weighted scorecard combining four signal categories:

Signal	Weight
Technical indicators (RSI, MACD, SMA crossovers, Bollinger Bands, momentum, volume)	40%
News sentiment (GDELT + Alpha Vantage polarity)	30%
Macro regime (VIX, 10Y–2Y yield spread, Fed Funds Rate)	20%
Sector strength (relative performance vs sector ETFs)	10%

The weighted score is mapped through a threshold to produce a direction. Confidence increases when multiple signals align.

v2 — Machine learning pipeline (WIP)

A supervised learning model designed to augment or replace the rules system. Features are stored as Parquet, built with scikit-learn, using historical signals as inputs and actual price direction as labels.

Planned improvements: nightly retraining, ensemble models, feature importance analysis, hyperparameter optimization, and rolling backtests.

Relevant modules: backend/app/services/prediction_engine.py and backend/app/services/ml_pipeline.py.

---

Architecture

Frontend (React)
      │
      ▼
Backend API (FastAPI)
      │
      ├── Data ingestion
      │     ├ Yahoo Finance
      │     ├ GDELT
      │     ├ FRED
      │     └ Alpha Vantage
      │
      ├── Prediction engine
      │     ├ Rules model (v1)
      │     └ ML pipeline (v2)
      │
      └── PostgreSQL
            └ Predictions + history

---

Project structure

.
├── backend/
│   ├── main.py
│   ├── Dockerfile
│   ├── requirements.txt
│   └── app/
│       ├── database.py
│       ├── api/
│       │   └── endpoints/
│       └── services/
│           ├── data_ingestion.py
│           ├── prediction_engine.py
│           └── ml_pipeline.py
│
├── frontend/
│   ├── public/
│   │   ├── index.html
│   │   └── favicon.svg
│   ├── src/
│   │   ├── App.js
│   │   ├── pages/
│   │   │   ├── TickerList.js
│   │   │   └── StockDetail.js
│   │   ├── components/
│   │   │   ├── Header.js
│   │   │   └── Footer.js
│   │   ├── utils/
│   │   │   ├── api.js
│   │   │   └── format.js
│   │   └── styles/
│   │       └── global.css
│   ├── package.json
│   └── vercel.json
│
├── docker-compose.yml
└── README.md

---

Local development

Prerequisites

• Python 3.11+
• Node 18+
• Git

Backend

cd backend
python -m venv venv
source venv/bin/activate    # Windows: venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env        # Edit with your API keys (optional)
uvicorn main:app --reload --port 8000

API docs available at http://localhost:8000/docs.

Frontend

cd frontend
npm install
npm start

Open http://localhost:3000.

Docker (full stack)

cp backend/.env.example backend/.env
docker compose up --build

---

Deployment

Backend → Railway

1. Deploy the /backend directory
2. Add a PostgreSQL plugin from the Railway dashboard
3. Set environment variables (ALPHA_VANTAGE_KEY, NEWS_API_KEY, FRED_API_KEY — all optional)

Frontend → Vercel

1. Deploy the /frontend directory
2. Set the environment variable:

REACT_APP_API_URL=https://your-railway-app.railway.app/api

---

Supported tickers

90+ instruments across multiple categories:

US Tech: AAPL, MSFT, NVDA, AMZN, META, TSLA, GOOGL, NFLX, AMD, INTC, QCOM, TXN, ORCL, CRM, ADBE, IBM

Financials: JPM, BAC, WFC, GS, MS, C, V, MA, BLK, PYPL

Healthcare: LLY, JNJ, UNH, PFE, MRNA, ABBV, MRK, TMO, GILD, REGN

Energy: XOM, CVX, SLB, HAL, MPC, PSX, VLO

ETFs: SPY, QQQ, IWM, DIA, VTI, GLD, SLV, TLT, HYG, XLK, XLV, XLF, XLE, XLY

International: TSM, BABA, TM, SONY, NVO, ASML, SAP, HSBC, BP, RIO, BHP, VALE

To add more tickers, edit GLOBAL_TICKERS in backend/app/services/data_ingestion.py.

---

Contributing

Pull requests are welcome. Areas where help is especially valuable: ML training improvements, backtesting, UI polish, alert systems, portfolio tracking, and additional ticker coverage.

git checkout -b feature/your-feature
git commit -m "Add your feature"
git push origin feature/your-feature

---

Acknowledgements

Built with FastAPI, React, Recharts, SQLAlchemy, yfinance, pandas, scikit-learn, APScheduler, GDELT, and FRED.

---

Support

If GPST helped you learn quant concepts or build trading ideas:

buymeacoffee.com/bedmakr

---

Built for the open-source and retail trading community. Not financial advice — use responsibly.