TWMA

This repository contains the pre-release code for the TWMA method as presented in our paper, "Enhancement of price trend trading strategies via image-induced importance weights."

Environment

Main Settings: Python 3.9 & Pytorch 1.11.0 & CUDA 10.2 & torchcam 0.3.2
Minor Settings: To be completed.

Data Pipeline

Script	Description
`build_image_dataset.py`	Plots stock price images and calculates labels.
`split_dataset.py`	Splits the built image dataset into training, validation and testing.

Network

Script	Description
`distributed_random_train.py`	Trains the ResNet "trader".
`dataset.py`	Defines the dataset structure based on PyTorch.
`distributed_utils.py`	Some useful functions for distributed learning.
`inference_(F)TGCN.py`	Obtains triple-I weights from the trained models.

Reproduce Part of Empirical Results

# Ensure you have updated the data path and log directory in each file.

# Step 1: Construct features and labels
python data_pipe/build_image_dataset.py
python data_pipe/split_dataset.py

# Step 2: Train trader.
CUDA_VISIBLE_DEVICES=0,1,2,3 python distributed_random_train.py

# Step 3: Inference and QCM learning
python network/inference.py

Cite

If you find this code helpful, please consider citing our paper:

To be completed.

Contact

Please feel free to raise an issue in this GitHub repository or email me if you have any questions or encounter any issues.