GitHub - Costasgk/ScoreCast: https://costas.pythonanywhere.com/ · GitHub
/" data-turbo-transient="true" />
Skip to content
Search or jump to...
Search code, repositories, users, issues, pull requests...
-->
Search
Clear
Search syntax tips
Provide feedback
--><br>We read every piece of feedback, and take your input very seriously.
Include my email address so I can be contacted
Cancel
Submit feedback
Saved searches
Use saved searches to filter your results more quickly
-->
Name
Query
To see all available qualifiers, see our documentation.
Cancel
Create saved search
Sign in
/;ref_cta:Sign up;ref_loc:header logged out"}"<br>Sign up
Appearance settings
Resetting focus
You signed in with another tab or window. Reload to refresh your session.<br>You signed out in another tab or window. Reload to refresh your session.<br>You switched accounts on another tab or window. Reload to refresh your session.
Dismiss alert
{{ message }}
Costasgk
ScoreCast
Public
Notifications<br>You must be signed in to change notification settings
Fork
Star<br>13
main
BranchesTags
Go to file
CodeOpen more actions menu
Folders and files<br>NameNameLast commit message<br>Last commit date<br>Latest commit
History<br>23 Commits<br>23 Commits
Scripts
Scripts
.gitignore
.gitignore
CONTRIBUTING.md
CONTRIBUTING.md
LICENSE
LICENSE
Procfile
Procfile
README.md
README.md
SESSION_NOTES.md
SESSION_NOTES.md
ScoreCast.png
ScoreCast.png
requirements-web.txt
requirements-web.txt
requirements.txt
requirements.txt
wsgi.py
wsgi.py
View all files
Repository files navigation
ScoreCast
A football match prediction web app powered by the Dixon-Coles Poisson model — covering 11 leagues across Europe, South America, and Asia.
🌐 Live at costas.pythonanywhere.com
What it does
ScoreCast scrapes historical match data from FBref, fits a Dixon-Coles model per league, and generates predictions for all upcoming fixtures. For each match it produces:
Win / Draw / Loss probabilities
Expected goals (xG) for each team
Full scoreline distribution (0-0 through 5-5)
Most likely score & top 3 scorelines
Both Teams to Score %
Over 1.5 / 2.5 / 3.5 goals %
Leagues covered
League<br>Country
Premier League<br>England
Serie A<br>Italy
La Liga<br>Spain
Ligue 1<br>France
Bundesliga<br>Germany
Super League<br>Greece
Serie A<br>Brazil
Serie B<br>Brazil
Eliteserien<br>Norway
Veikkausliiga<br>Finland
J1 League<br>Japan
Features
Best Picks — fixtures where the model has 65%+ confidence, grouped by date
Simulator — pick any two teams from any league and run a custom matchup
Rankings — model strength ranking vs actual league standings, with over/underperformer highlights
Team DNA — attack/defence ratings, win %, goals scored/conceded, last 5 form
Model Accuracy — backtested accuracy across all leagues for the last 12 months
Visitor Stats — lightweight analytics dashboard at /stats
How it works
Scrapping.py → Cleaning.py → ScorelineModel.py → Flask web app<br>FBref clean & Dixon-Coles serves<br>scraper validate Poisson fit predictions
Scraping — undetected_chromedriver bypasses Cloudflare on FBref; incremental saves per season; persistent Chrome profile so Cloudflare only needs solving once
Cleaning — normalises columns, filters bad rows, exports per-league CSVs
Modelling — fits attack/defence parameters per team via maximum likelihood (L-BFGS-B); exponential time-decay (half-life ~107 days) weights recent matches far more than old ones; Dixon-Coles low-score correction adjusts 0-0, 1-0, 0-1, 1-1 probabilities
Pipeline — pipeline.py orchestrates all three steps with smart staleness detection
Running locally
# 1. Clone and set up environment<br>git clone https://github.com/Costasgk/ScoreCast.git<br>cd ScoreCast<br>python -m venv env<br>env\Scripts\activate # Windows<br>pip install -r requirements.txt
# 2. Run the pipeline (scrape → clean → predict)<br>cd Scripts<br>python pipeline.py
# 3. Start the web app<br>cd WebApp<br>python app.py
The app runs at http://127.0.0.1:5000.
Note: First-time scraping takes several hours (FBref rate limits). Subsequent runs are incremental and much faster.
Project structure
Scripts/<br>├── Scrapping.py # FBref scraper<br>├── Cleaning.py # data cleaning<br>├── ScorelineModel.py # Dixon-Coles model<br>├── pipeline.py # orchestrates scrape → clean → predict<br>└── WebApp/<br>├── app.py # Flask app<br>├── templates/ # HTML templates<br>└── static/ # CSS, fonts, favicon
Datasets/<br>├── Scrapped Datasets/ # raw FBref output<br>├── Cleaned Datasets/ # cleaned per-league CSVs<br>├── Predictions/ # upcoming fixture predictions<br>└── Models/ # fitted model parameters (JSON)
Deployment
Hosted on PythonAnywhere (free tier). To redeploy after regenerating predictions:
Rebuild the deployment zip locally
Upload to PythonAnywhere Files tab
Unzip and hit Reload on the Web tab
License
MIT — free to use, modify, and...