/dfspy

Daily fantasy sports lineup optimization package for Python.

Primary LanguagePythonMIT LicenseMIT

dfspy

Python-Versions LICENSE

dfspy is a Daily Fantasy Sports lineup optimization package for Python.
The main optimization algorithm scrapes current data from web sources, and uses machine learning algorithms and convex optimization to return optimal lineups for the current NFL week.

Table of contents

Installation

Install Repo

From terminal:

git clone https://github.com/jason-r-becker/dfspy.git

Set up venv

Using Anaconda, from terminal:

cd dfspy/
conda create -n dfspy python=3.7
source activate dfspy
pip install -U pip
pip install -r requirements.txt

Scraping Data

Directory Organization

The data directory organization is shown below, structured <season year>/<week>/<player position>/<source>.csv. Source name files contain projections from said source whereas STATS.csv contains the true realized stats. Week 0 refers to full season projections/stats.

data
├── 2016
├── 2017
└── 2018
    ├── 0
    ├── 1
    ├── ...
    └── 17
        ├── DST
        ├── K
        ├── QB
        ├── RB
        ├── TE
        └── WR
            ├── CBS.csv
            ├── ESPN.csv
            ├── FFToday.csv
            ├── FantasyPros.csv
            ├── NFL.csv
            ├── RTSports.csv
            ├── STATS.csv
            └── Yahoo.csv

Scraping

Historical projections and stats as well as current projections can be scraped with the scrape_data.py module. The following command line options are used to specify scraping parameters.

Setting Command Line Keyword Default
Sources -s, --sources All Projection Sources
Week(s) of the season -w, --weeks Current Week
Season Year(s) -y, --years Current Season

For example, scraping projections for the current week can be accomplished:

python scrape_data.py

To specify historical projections to scrape, command line options can be used. To scrape full season projections from 2018:

python scrape_data.py -w 0 -y 2018

Similarly for all individual weeks (or specified weeks):

python scrape_data.py -w 1-17 -y 2018

All data (full season and weekly) for given years can also be scaped:

python scrape_data.py -w all -y 2016-2018

Finally, true realized stats can be scraped by specifying the source. Similarly any individual source can be scraped.

python scrape_data.py -w all -y 2016-2018 -s STATS

Contributions

GitHub-Commits GitHub-Issues GitHub-PRs

All source code is hosted on GitHub. Contributions are welcome.

LICENSE

Open Source (OSI approved): LICENSE

Authors

The main developer(s):