VIX Direction Prediction with LSTM-Based Deep Learning
Francisco Luis de Avelar Cardoso
Published
0
A full deep learning pipeline for predicting the daily directional movement of the VIX volatility index.
The system uses:
Engineered features from SPY, QQQ, and VIX
A PyTorch LSTM backbone with an MLP classifier head
Rolling-window time series cross-validation to avoid leakage
Includes an interactive Streamlit dashboard for model diagnostics, fold-level analysis, and configuration comparison, showcasing realistic validation practices in financial ML.