Non‐stationary Structure in High‐frequency Market

Purpose
Identify the non‐stationary correlation structure and volatility in cryptomarket and Forecast the asset price.
Method
- Model based: (near)Unit‐root asymptotical inference
- Machine Learning: Neural Network time series prediction
- RNN-LSTM
Role
Solo research and project. Managed at github.com/monchewharry/Crypto_Forecasting_kaggle.