📈 Stock_Prediction_ML — AI‑Driven Stock Movement Forecasting

Author: Sam Ginzburg
Email: samginzee@gmail.com
Tech Stack: Python · Pandas · XGBoost · Streamlit · SimFin API · Matplotlib

---

Business Problem & Motivation

Retail investors and analysts are often inundated with market data but lack simple, interpretable tools that combine financial fundamentals and technical indicators into actionable insights.

The goal of Stock_Prediction_ML is to demonstrate how a data analyst can:

• Ingest and clean real‑world financial data
• Engineer meaningful, explainable features
• Apply machine‑learning models appropriately
• Deploy insights through an interactive application

Rather than attempting to "beat the market," this project focuses on directional forecasting and decision support using a concise, well‑defined feature set.

---

Project Scope

This project delivers an end‑to‑end stock analysis and prediction application that:

• Pulls live and historical stock data via API
• Merges daily price data with quarterly financial statements
• Engineers valuation and trend‑based features
• Applies a trained XGBoost model to generate Buy / Sell signals
• Presents results through a clean, interactive Streamlit interface

The application is intentionally scoped as a forecasting and analytics prototype, not an automated trading system.

---

Methods / Skills Demonstrated

This project showcases the following technical and analytical skills:

• Python (data processing, modeling, application logic)
• Pandas (data cleaning, joins, time‑series transformations)
• API Integration (SimFin financial data)
• Feature Engineering (valuation and technical indicators)
• Machine Learning (XGBoost classification)
• Model Serialization & Loading
• Streamlit (interactive analytics dashboard)
• Data Visualization (Matplotlib)
• Environment Variable Management (.env files)
• Production‑minded Project Structuring

---

Features Used in the Model

After exploratory analysis and feature testing, the final model uses a concise and interpretable feature set:

Feature	Description
Closing Price	Captures short‑term market behavior
P/E Ratio	Fundamental valuation signal
50‑Day SMA	Medium‑term trend indicator

The target variable is next‑day price movement direction (up/down).

---

Results

The application successfully:

• Displays merged historical financial and price data
• Visualizes one‑year closing price trends
• Generates real‑time Buy / Sell signals for selected stocks
• Handles non‑trading days and missing data gracefully

Predictions are presented as directional signals, prioritizing interpretability over unrealistic accuracy claims.

---

Quick Glance at the Application

Below is a snapshot of the Streamlit application, highlighting the core analytical functionality and user experience.

Stock Selection Interface

Users can select from major U.S. large‑cap equities (the “Magnificent Seven”) using a simple sidebar control.

---

Historical Price Trend Visualization

The application visualizes the selected stock’s closing price over the past year, enabling quick trend inspection and context before viewing model output.

---

Machine Learning Prediction Output

Based on the most recent available trading day, the trained XGBoost model generates a directional Buy / Sell signal using engineered financial and technical features.

---

Project Structure

Stock_Prediction_ML/
│
├── app_new_structure.py    # Main Streamlit application
├── pages/
│   └── go_live.py          # Live prediction interface
├── notebooks/
│   ├── training.ipynb      # Model training & experimentation
│   └── kpi2.ipynb          # Feature analysis
├── data/
│   └── mag7_data.zip       # Processed datasets
├── assets/
│   └── mag7_data.zip  
├── mag7_final_model.json   # Trained XGBoost model
├── simfin_api.py           # Custom API wrapper
├── requirements.txt
├── .gitignore
└── README.md

---

How to Run This Project

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt

3. Create an environment file:

   cp keys.env.example -> keys.env

4. Inside the 'keys.env' file, please insert the following information:

   SIMFIN_API_KEY="{insert your API key here}"

5. Launch the application:

   streamlit run app_new_structure.py

The keys.env file is excluded from version control.

---

Limitations

• Predictions are short‑term and directional only
• External macroeconomic events are not modeled
• This project is designed for learning and demonstration purposes

---

Disclaimer

This project is for educational and analytical purposes only and does not constitute financial or investment advice.