Intraday Volatility Prediction Model

This repository contains a Python implementation of intraday volatility prediction inspired by the paper A Practical Model for Prediction of Intraday Volatility from Young Li. The project extends the paper by integrating GARCH models with diurnal patterns to enhance intraday volatility forecasting for selected stocks.

Features

• Data Pipeline: Retrieves minute, hourly, and daily stock data from a SQL database.
• Volatility Models:
  • EWMA (Exponentially Weighted Moving Average) volatility model.
  • GARCH(1,1) model, with flexibility to use other GARCH variations.
• Diurnal Profile: Estimates intraday volatility using the Garman-Klass volatility estimator.
• Intraday Volatility Prediction: Scales daily volatility to intraday levels with diurnal components.
• Visualization: Provides functions to visualize predicted vs. realized volatility and diurnal patterns.

---

Architecture

The key components of this project are:

1. Data Retrieval

• Pipeline: Class for fetching stock data at different frequencies (minute, hour, day).

2. Volatility Models

• ewma_vol(): Calculates daily volatility using the EWMA method.
• garch_vol(): Computes daily volatility using the GARCH(1,1) model, with options for variations.

3. Diurnal Profile Calculation

• Utilizes Garman-Klass volatility to estimate normalized intraday volatility for different times of the day.

4. Intraday Volatility Prediction

• Combines daily volatility, diurnal components, and dynamic averages to predict future intraday volatility: