jasminesor / dmr

A Demo for "Distributed Mean Reversion Online Portfolio Strategy with Stock Network"

DMR strategy demo

This is a project demo of the DMR strategy, which helps users to analyze the performance of a universal strategy based on the mean reversion assumption - the DMR strategy - compared to other universal strategies.

Please note: If you use any part of the code in this repository for your own work, we kindly request that you cite our paper. Citing our work helps us to receive credit for our contributions and encourages further collaboration. You can find the citation information for our paper below:

• Title: Distributed mean reversion online portfolio strategy with stock network
• Authors: Yannan Zhong, Weijun Xu, Hongyi Li, Weiwei Zhong
• Journal: European Journal of Operational Research
• DOI Link: https://doi.org/10.1016/j.ejor.2023.11.021

Thank you for respecting our work and contributing to the open source community!

Preparation

To run the demo properly, your computer needs to meet the following requirements:

• Python 3.8[important]: Please make sure you have Python installed. The first recommendation is to use python 3.8, higher versions are theoretically possible.

• BuildTools[optional]: If your system is Linux or MacOS, you do not need to pay attention to this. Also, please ensure that the system account you are using has write permissions when you running this software.

  If your computer system is Windows, please run "vs_BuildTools.exe" in our project file to generate the underlying build code needed for "cxvopt" package in python.

  The following screen may appear when you open it, please wait for the program to run automatically.

  

  Once successfully run, please check the appropriate boxes as shown below, then click "Install" in the bottom right corner and wait for it to finish.

  

Introduction

RunMe.py

You can run the python file to get the comparison and analysis results of all the strategies. Enter python .\RunMe.py --help in your cmd or powershell console to get the usage instructions.

DMRdemo> python .\RunMe.py --help
	
usage: RunMe.py [-h] [--r R] [--d D]
	
optional arguments:
	  -h, --help  show this help message and exit
	  --r R       result class: performance, heatmaps, heatmaps_alpha. for example: --r='performance' will calculate the
	              performance of the chased strategies.
	  --d D       dataset: run strategies in the dataset you select. default is 'all'.for example: --d='name' will run the
	              strategies in 'name.xlsx', --d='name1,name2' will run strategies in 'name1,xlsx' and 'name2.xlsx'.
	              Attention: the .xlsx files must in './data' folder before you run it!!

• --r=[resultname]: chose the result: "performance", "heatmaps", "heatmaps_alpha". default is "performance".

  • performance: strategies performance, contains "total_wealth", "sharpe ratio", "information ratio", "alpha", "beta", "annualized_return", "annualized_volatility", "drawdown_period", "max_drawdown", "annual_turnover", "mean_excess_return", "calmar ratio", "run_time".

  • heatmaps: calculate the heat maps of DMR strategy in all the dataset.

  • heatmaps_alpha: calculate the heat maps of alpha_k of DMR strategy in all the dataset.

    for example: --r="performance" will calculate the performance of all the strategies.

• --d=[datasetname]: run strategies in the dataset you select. default is 'all'.

  for example: --d='name' will run the strategies in 'name.xlsx', --d='name1,name2' will run strategies in 'name1,xlsx' and 'name2.xlsx'

  Attention: the .xlsx file must be placed in the ./data folder in advance!

data

This folder contains your dataset file, which must be an excel file in .xlsx format. The first row of the dataset file should be the stock name, and from the second row onwards, the stock closing price in ascending chronological order. for examples:

date	stock 1	stock 2	$\cdots$	stock m
day 1	price 1	price 1	$\cdots$	price 1
day 2	price 2	price 2	$\cdots$	price 2
$\vdots$	$\vdots$	$\vdots$	$\vdots$	$\vdots$

result

This folder contains all kinds of results. If performance is selected when running RunMe.py, a ./result/performance folder will be created to store the results. If heatmaps is selected, a ./result/heatmaps folder will be created to store the results. If heatmaps_alpha is selected, a ./result/heatmaps_for_alpha folder will be created to store the results.

Quick Start

To ensure that you have all the third-party libraries required for your project installed in python, please run the following command before starting your project

pip install -r requirements.txt

The following commands will help you to get started quickly with this project.

Running 'performance' of strategies in dataset 'msci' and 'FS100'

The following two commands have the same effect.

python ./RunMe.py --r="performance" --d="msci,FS100"

or

python ./RunMe.py --r="performance" --d="msci, FS100"

This command will run the performance results of all strategies on the msci dataset. The results are saved in the ./results/performance folder.

Running 'heatmaps' of DMR strategy in all the dataset

The following two commands have the same effect.

python ./RunMe.py --r="heatmaps" --d='all'

or

python ./RunMe.py --r="heatmaps"

This command will include a heat map of DMR strategy in all the datasets placed in the ./data folder. The results are saved in ./result/heatmaps.

Running 'heatmaps_alpha' of DMR strategy in dataset 'msci'

python ./RunMe.py --r="heatmaps_alpha" --d="msci"

This command will test the cumulative wealth of DMR strategies with different alpha_k on the 'msci' dataset.

Disclaimer

This software is only used for verification and exchange of learning purposes. And it is far from any production environment. Do not risk money which you are afraid to lose. Use the software at your own risk. The authors assume no responsibility for your trading results.