GitHub Projects Documentation

Realized Volatility Prediction (realized-vol-prediction)

Overview

Building a predictive model for stock volatility. Main code in ml_my_beloved.ipynb.

Key Features

• Python libraries (Pandas, Numpy) for data manipulation.
• Machine learning models for stock market volatility prediction.
• Data preprocessing, feature engineering, and model evaluation.

Technical Summary

1. Data Preparation: Loading and preprocessing stock market data.
2. Feature Engineering: Calculating WAP, log returns, and realized volatility.
3. Model Building: Predicting stock market volatility using XGBoost.
4. Evaluation: Metrics like R2 score and RMSPE.

Squared Growth Expectation Model (Squared growth expectation model.ipynb)

Overview

Modeling squared growth expectation of financial instruments.

Key Components

• Data manipulation using Pandas and Numpy.
• Analysis of order book data.
• Calculation of LWAP and LDIFF.

Technical Details

1. Data Loading: Reading financial data.
2. Feature Calculation: LWAP and LDIFF calculations.
3. Statistical Analysis: Modeling squared growth expectation.

Exponential Weighted Moving Average - Version 2 (ewma-v2.ipynb)

Overview

Improving stock volatility predictions using EWMA.

Features

• Advanced data processing (Pandas, Numpy).
• EWMA implementation for volatility predictions.
• Model performance evaluation (R2 score, RMSPE).

Technical Aspects

1. Data Preparation: Handling financial time series data.
2. EWMA Implementation: Enhancing predictive models.
3. Model Performance: Analyzing effectiveness using statistical metrics.

Machine Learning Model for Volatility Prediction (ml_my_beloved.ipynb)

Overview

Predicting stock market volatility using XGBoost.

Highlights

• Data preprocessing and feature engineering.
• XGBoost regressor implementation.
• Performance evaluation (MAE, MSE, RMSE, MAPE).

Key Processes

1. Data Engineering: Feature creation (WAP, bid-ask spread, order imbalance).
2. Model Training: XGBoost model tuning and training.
3. Evaluation: Accuracy and error metrics assessment.