Final Project

This is the Final Project for DTFF: Risk Parity Model based on MVT

Teammates

Name	Student Number
Zifeng Tang	21-742-093
Haoran Zhu	21-742-069
Yuzhi Mao	21-742-218

Database

Database Present

---

├── data
│   ├── raw     (out)  ---> python and matlab codes would read data from this folder, the data we offered in .ftr and .xlsx form.
│   ├── result  (out)  ---> This datas would be used in Rshiny
│          ├── matlab    (in)  <--- The final data given by matlab codes  
│          └── python    (in)  <--- The final data given by python codes

---

Remarks

1. The folder data has two folders raw and result, (out) means we get datas from the database and (in) means we put the datas into the database.
2. In order to be convenient for the users, python and matlab codes in the folder \scr and r code in the folder \app would get or put datas with the folder \data, which means it is a small database.
3. For the user who wants to run the python code in docker, we set a independent database in the folder docker\python because we don't want to copy the \data folder in the root directory to docker container.
4. In order to try new data types, we read in feather type data in the python code.
5. The raw data "Original_Closing_Price" is got from choice software in china.

FILEPATH

In order to avoid using absolute addresses, we define a path variable FILEPATH to denote the Root directory for this project. However, you don't need to change it when using any codes in \scr and \app folders if you

1. Run python in VScode directly: we used os library to solve it;
2. Run matlab in Matlab: we used cd(..) and CURRENT_PATH=pwd to solve it;
3. Run r in Rstudio: we used ..\ when reading datas;

If you want to clone the repo to VS Code directly, you need to change the FILEPATH by instructions in the codes 1 .\scr\python\main.py; 2 .\app\app.R;

eg1: for `main.py`
# If you run the code in your own IDE
FILEPATH=os.path.abspath(os.path.join(os.getcwd(),"../.."))
# If you run the code in VSCode in the entire repo: I mean clone the all repo into VScode
FILEPATH=os.path.abspath(os.path.join(os.getcwd()))

eg2: for `app.r`
# If you run R code in VScodes in the entire repo: I mean clone the all repo into VScode
   DATAPATH <- "data/"
# Otherwise if you run rcodes in rstudio
   DATAPATH <- "../data/"

But we do not recommend you to do this, because you may need to do some other operations to adapt to the Vscode environment to achieve the purpose of running .py, .tex, .r three types of files in the same project. However, we also export the environment as Digital_Tools_for_Finance.yml .

Moreover, if you indeed want to change the path of root directory, what you need to do can be listed as

1. Python codes: you only need to change the FILEPATH in the main.py;
2. MATLAB codes: you only need to change the FILEPATH in the main_function.m;
3. r codes: you only need to change the FILEPATH in the app.r;

Codes and Environments

We offer two types of codes for this project: Python and Matlab, you can run all of them using your IDE and the environment setting would be given later. For Python code, we also offer a way to run it in docker, while we only offer a simple dockerfile for matlab code since you need to update some toolboxes by yourself.

1.Python Codes

We use the python code to implement the risk parity model in our report, while it would only give you the results of the model which include:

1. The weights of assets in the model; (.xlsx)
2. The net value of portfolio based on different models; (.xlsx) and all of figures in the report are drawn by MATLAB code.

All of the python codes are given in the folder src\python, you can run it in the ways

1. Your own IDE and install all the packages in the file requirements.txt.
2. We also give you the environment export Digital_Tools_for_Finance.yml , you can run it in vscode under conda environment.
3. Run it in docker. If you want to do this, please see the folder docker\python, what you need to do is
   1. Go in the folder docker\python using command Lines;
   2. Open the docker and run the command line:

   docker build --tag dtff-python .
   

   and

   docker run -it -v "$(pwd)/out:/out" dtff-python
   

   3. Then you would see all of the results are shown in the folder docker\python\out.

2.Matlab Codes

We use the matlab codes to implement the same model as Python code and draw all of the figures in reports. In order to run the code, you only need to do

open the main_function.m in matlab -> install the toolbox Optimization Toolbox and Parallel Computing Toolbox-> Run main_function.m .

All of results would be exported to the folder data\result\matlab directly and all of the figures would be shown on your computer at the same time.

Note: if you run the codes at first time, please write the function parpool(n) in the main_function.m to set your parallel computing numbers, n must be smaller than the kernels of your computer.

3.Important Hints

Since we used parallel computing in Matlab codes, so it would be much faster than the python codes, it is normal that the python code takes a very long time to run. Therefore, we would give you a flag to show that the code runs normally such as

```
Model based on Std Start:
Process: 5%
Process: 10%
...
```

Report and Slides

The report of our final assignment is called DTFF_report_final.pdf and the LaTeX-interpretable code and the used figures are included in the folder reports.

The slides of our final assignment is called DTFF_Presentation.pdf and the LaTeX-interpretable code and the used figures are included in the folder slides.

Interactive App Based on R Shiny

We create an interactive app using R shiny whose code is in app/app.R to show the plots of the daily return, the assets correlation coeffient, weights, and the net values in general cases. Specifically, we can determine Output to show the different kinds of plots and choose Start Date and End Date to determine the date interval. And all packages have been listed in the requirements.txt.

If you don't want to run by yourself, you can use the link to get the result on website shinyapps.io:

DTFF-Final-Project-Rshiny-results

Project Organization

---

├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── raw       <- python and matlab codes would read data from this folder, the data we offered in .feather and \.xlsx form.
│   ├── result      
│          ├── matlab      <- The final data given by matlab codes
│          └── python            <- The final data given by python codes
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as LaTeX
│   └── figures        <- Generated graphics and figures to be used in reporting
│
│
├── setup.py           <- makes project pip installable (pip install -e .) so src can be imported
├── src──                <- Source code for use in this project.
│   ├── matlab 
|       ├──data     <- The codes used to read datas
|       |  
│       ├──features <- The codes used to calculate the features that used in model
|       |
|       ├──model    <- Risk_parity model
|       |
|       ├──visualization <- Plot the results
|       |
│   ├── python
|       └── __init__.py    <- Makes src a Python module
|       |  
|       ├── main.py <- The main function to run all the codes
|       |
|       ├── data <- Scripts to download or generate data
│       |   ├── MakeData.py
|       |   ├── downstream.py
|       |
|       ├── model
            └── Risk_Parity.py

---