EvoFolio

More information is available in the article: EvoFolio - Research Article

Table of Contents

• Description
• Installation
• Run
• Parameters

Description

EvoFolio is a project that implements the NSGA-II algorithm applied to portfolio optimization. It computes different financial metrics (like yield and risk) based on stock data, and allows visualization of the generated data.

Installation

The latest version can be cloned with the following command:

git clone --recurse-submodules https://github.com/Balbus95/EvoFolio.git

Create a Virtual Environment (Optional but Recommended):

Use the following command to create a virtual environment:

python3 -m venv venv

Note: to activate and deactivate the virtual environment, use source venv/bin/activate and deactivate on Linux/MacOS, or venv\Scripts\activate and deactivate on Windows.

Install Dependencies

Dependencies, including the custom DEAP submodule, can be installed by running:

pip install -r requirements.txt

Notes on manual installation of the custom DEAP submodule

In case you need to install the custom DEAP library manually, use one of the following commands:

pip install .\deap --use-pep517

Or:

pip install ./deap

Run

To run the main optimization script:

python "final hybrid.py"

Other scripts

You can run other utility scripts in the same way:

• loadfile.py: Run this script to open a Tkinter GUI to view and plot graphs of the .dump files created by the main script. It supports both monthly and trimestral outputs.
• stockToPDF.py: Run this script to generate PDFs of the trend graphs (close prices) for each stock. PDFs will be saved in the stockToPDF_out folder.

Parameters

Default Parameters of the Main Script (final hybrid.py)

The following parameters can be modified directly at the beginning of final hybrid.py:

MINAZIONI, MAXAZIONI= 10, 14 # min and max number of different stocks that a portfolio can hold
BUDG = 1000000 # initial budget of portfolios (USD$)
BOUND_LOW, BOUND_UP = 0, BUDG # min and max number of equal stock that a portfolio can hold
NDIM = len(stockdf) # portfolio size (number of stock's files)
SELPARAM= 0.8 # NSGA-II selection parameter, e.g. 0.8 selects 80% of the pop
CXPB = 0.9 # probability of mating each individual at each generation 
ELITECXPB=0.3 # probability of mating with an elite
MAXTIME=24 # maximum csv row to read, the row is the date in the csv

Note: the parameters below might be overwritten by a loop in the script. To change them, modify the loop definitions near the bottom of final hybrid.py:

MU = 250 # population size, number of individuals in the population.
TOURNPARAM= 0.9 # tournament parameter, e.g. 0.9 selects 90% of the pop
NGEN = 50 # number of generation of nsga2

Default Paths

The input and output folder paths are set by default to work seamlessly on both Windows and Unix systems. If you need to change them, look for PATHCSVFOLDER, PATHLOGBMONFOLDER, etc., in the scripts:

• Inputs: Inside final hybrid.py, loadfile.py and stockToPDF.py. Default stock data path is ./stock/WEEK.
• Outputs: Handled in final hybrid.py and loadfile.py. Default output paths are ./output/mensile/... and ./output/trimestrale/....

---

☕ Support

If my work was useful, a coffee is always welcome!