Change in portfolio turnover calculation and test, Ledoit-Wolfe cov, base for Github actions #14

huuphuocle · 2024-12-09T07:01:34Z

Hello,

Here are the details of my PR:

Some changes in portfolio.py: correct calculations, some perf improvement through binary search in portfolio list
Ledoit-Wolfe covariance: sklearn Ledoit-Wolfe shrinks only towards identity matrix it seems
some setup for a Github actions, a test class for portfolio.py and some changes in test_quadratic_program.py.
a minor fix in data_loader (because of the data format change in this PR: ✨ Standardize CSV File Formats for Consistency (#8) #10

Please let me know your feedback on this. Thanks

…ata_loader. Add requirements.txt, change GitHubActions yml file

…t are not in current selection

huuphuocle and others added 13 commits November 21, 2024 06:18

Merge branch 'feature_adaptive_selection'

2c20137

Fix(Data) Adapt data_loader to new format

6f836ec

Create run-tests.yml

0d4449f

Improve(CI/CD) Test qp: load data only once, adapt after changes in d…

c9bef72

…ata_loader. Add requirements.txt, change GitHubActions yml file

Fix(CI/CD) Add highs to requirement, remove python 3.9 from yml config

8fc4d15

Feature: Ledoit Wolf covariance

fc268d2

Correct some calculations in portfolio, use binary search for querying

1114b9d

Merge branch 'GeomScale:main' into main

3f12249

Test portfolios

1ef0760

Merge remote-tracking branch 'origin/main'

dc27ad8

Use MSCI data for test_portfolio

8c98dcb

Handle init weight in backtest

d04ce48

add builder for turnover constraint, handle assets in prev weight tha…

d014534

…t are not in current selection

Provide feedback