Street Fighter 6 Bracket Predictor

This project aims to predict tournament results for Street Fighter 6 using historical match data from Street Fighter V and 6.

How it works

• Data Extraction: Extract player, event, and match result data for both Street Fighter V and VI from start.gg within API limits using extract_startgg_data.py.
• Data Scraping: Scrape additional data from Liquipedia using a manually compiled CSV file containing tournament names and URLs with scrape_liquidpedia.py.
• Data Collation: Extract more detailed player data from start.gg and merge all extracted and scraped data into consolidated tables using collate_data.py.
• ELO Calculation: Compute ELO ratings and tournament participation statistics for every player for each event they participated in using calc_elo.py.
• Model Training: Train a machine learning model to predict match outcomes based on the data gathered and processed in previous steps using model_training.ipynb.

Install

Prerequisites

Before installation, ensure you have Python 3.11 installed on your system. You can download Python here.

Steps

1. Clone the Repository: Clone this repository to your local machine using:

git clone https://github.com/SarahEGood/sf6predictor

2. Environment Setup: Set up a virtual environment:

python -m venv sf6-predictor-env
source sf6-predictor-env/bin/activate # On Windows use `sf6-predictor-env\Scripts\activate`

3. Install Dependencies: Install the required Python libraries:

pip install -r requirements.txt

4. API Token: Before you can extract any start.gg data from the start.gg API, you will need an API token. Follow the instructions here to obtain a start.gg authentication token. Set this token as an environment variable:

export startgg_token='YOUR_TOKEN_HERE'  # On Windows use `set startgg_token=YOUR_TOKEN_HERE`

5. Setup Liquipedia Webscraping (Optional): If you wish to also use Liquipedia tournaments, you will need to manually create a file called scrape_brackets.csv to include the urls and other data you want to scrape from Liquipedia. Please note that the current scraping code supports individual tournaments only; group events are not supported. Ensure that the file includes the following columns:

• event_id: An integer less than 1000, unique for each event. Reuse this ID if an event has multiple pages for scraping.
• url: The URL to the page that needs to be scraped.
• event_name: The name of tournament.
• comptier: The competitive tier assigned by Liquipedia for the tournament.
• func_type: An integer that determines which function to use.
  • If only brackets are displayed, set func_type to 1.
  • If pools are displayed with expandable match data, set func_type to 2.
  • If aggregate pools data is displayes with no specific match data (i.e. you know who won the pool but not specifically who they matched against), use func_type 3.
• date: The date of tournament's last day.
• country: The country of tournament (optional)
• city: The city of tournament (optional)
• state: The state or province of tournament. Should follow ANSI 2-letter code. (optional)

This is an example setup for this csv:

event_id	url	event_name	comptier	func_type	date	country	city	state
0	https://liquipedia.net/fighters/Abang_Bald_Cup/5/Group_Stage	Abang Bald Cup #5	2	1	2024-28-02
0	https://liquipedia.net/fighters/Abang_Bald_Cup/5/Bracket	Abang Bald Cup #5	2	1	2024-28-02
1	https://liquipedia.net/fighters/Keio_Cup/2024/Bracket	Keio Cup	2	1	2024-28-01	JP	Tokyo
2	https://liquipedia.net/fighters/Gamers8/2023/SF6/Group_Stage	Gamers 8 2023	1	2	2023-09-13	SA	Riyadh
2	https://liquipedia.net/fighters/Gamers8/2023/SF6/Playoffs	Gamers 8 2023	1	1	2023-09-13	SA	Riyadh

Usage

To run the full sequence from data extraction to predictions, run the following from the project directory:

python extract_startgg_data.py
python collate_data.py
python scrape_liquidpedia.py # Only run if scrape_brackets.csv is setup
python calc_elo.py
jupyter notebook model_training.ipynb

License

The Street Fighter 6 Bracket Predictor is available under the MIT license. See the LICENSE file for more info.