Stock Prediction Framework

This framework is built upon multi-GPU DDP (Distributed Data Parallel), utilizing minute-level factors to predict stock price movements.

Algorithms

There are already four algorithms in this framwork (Vanilla, iTransformer, deeplob, GRUModel), you can use different algorithms in configs/config.py

Sorry, due to confidentiality constraints, I am unable to provide the most efficacious algorithm.

Vanilla

traditional transformer structure

GRU

GRU framework model

Data

Sorry we cannot provide any data, because the data we use have private factors

We utilize minute-level stock data, employing various factors as features, to predict the rise and fall of stock prices. You can make your own data.

Training Framework

Leveraging Distributed Data Parallel (DDP) Architecture for Model Training