Football/Soccer Match PREDICTION MODEL doesn't work | Poisson Distribution in Football

This video features the power and limitations of the Poisson model, a statistical tool that used often to predict football/soccer matches and goals/score.
While the Poisson distribution is a handy tool in many areas, it falls short when it comes to predicting goals in football.
Support my channel by buying me a virtual coffee on www.buymeacoff...
Your support will help me continue to provide you with quality insights, tutorials, and resources for the world of football analytics.
► Getting Started with Football Data Analysis Using Python: www.buymeacoff...
Contents:
0:45 Introduction to Football/Soccer Goals Prediction with the Poisson Model
01:49 What is the Poisson Distribution?
02:54 Why Poisson Prediction Model doesn't work?
03:14 Football goals are not independent and don't follow the Poisson Distribution
03:43 The Unpredictable Nature of Football
04:19 Football Team Performance Matters
04:47 Star Players and Injuries
05:12 High-Scoring and Low-Scoring Matches
05:40 Football Has Too Many Variables
06:01 Predictions with Poisson Distribution in Football summary
Useful links:
- Football/Soccer Goals PREDICTION with the Poisson Model | Poisson Probability in Football
• Football/Soccer Goals ...
- Getting Started with Football Data Analysis Using Python:
• Football/Soccer Data A...
Buy me a virtual coffee: www.buymeacoff...

[@gilsonjunior_narrador]

Ok, I agree with you. But what can we do to improve this prediction ? Include available players overall strength for both teams ? Include a coef like elo rating ? Include defense and attack overall strength ? to separate all strengths in home and away ? Regards.

[@ciao.football]

The way I see it, it depends on what's your goal and what you want to accomplish with your predictions. If it's for betting, I'd say this model can be improved with few tweaks, adding maybe current ranking, last season position, squad transfer cost and things like this. That would be probably enough, because in betting models it's not only the prediction that matters, but also money & risk management (and I'd say it's even more important)

[@gilsonjunior_narrador]

I just want to know the probability of championship winner....

[@ciao.football]

So in this case I'd say weighing current positions is something to add to the model. Transfer value of squads as well (it has ~90% correlation with final positions in a championship I've tested).

[@ciao.football]

Check out what the Poisson Distribution is in more detail here: kzbin.info/www/bejne/iqG4hpaOft51j9k