World Premiere – Expected Goals in Sweden, Machine Learning and the Blog!

Hello World!

So this is my first blog post. There will be a lot more to come! I’ll probably mix stories and insights from my ongoing projects regarding football analytics, predictions and betting with Team/Player/Match-Analysis. The latter actually has sprung from the projects so all off a sudden I sat with what I think is a world premiere of projecting Expected Goals for the top Swedish divisions.

I also think that the way I project ExpG is somewhat different from how e.g. Michael Caley makes his. My approach, being more of a programmer than statistician has been very pragmatic. I took all the data I could find about finishes; position on the pitch, game time, possession in 5 min interval and whether the finish is from open play, set piece or from a corner. I then use Machine Learning to make the projection. I haven’t seen anyone doing this, a bit strange since I find it to be quite a good match. I will try to do some verifications of the model in the future.

More on how I got here will come in blog posts during the fall. It is quite a long story. So now on to yesterdays games…

Mff – Örebro

Mff obviously dissapointed. Still Örebro had a few good chances.

Åtvidaberg – Gefle

Typical draw with these two teams involved 🙂

Halmstad – Kalmar

Huge dissapointment for Kalmar. They created enough chances to win this one.

Oh, one more world première I think, is visualizing the match events using Plotly. Quite nice isn’t it? I had to draw the lines using formulas but other from that it adds nice interactivity with player names on hover as well as the full screen option.

Anyone interested in some specific game, player or team in Sweden? I’m up for requests!

Published by

Ola Lidmark Eriksson

Football analyst/programmer

3 thoughts on “World Premiere – Expected Goals in Sweden, Machine Learning and the Blog!”

  1. hi!

    my name is alex and im from germany. i just started to build my own expG models but i want to go deeper into prognosis of future games

    i found online that using support vector machines (or basically machine learning) would be a good way to do that
    but i have NO clue how they work^^

    so i thought maybe you could explain to me what you are doing exactly and which data you use for your predictions using machine learning? you are the first person i found you uses this method for soccer and i think this might be the right way to go and want to learn about it

    well…you have my email, so… 🙂

    thanks in advamce


    1. Hi!
      Thanks for reading! Predicting future games is a whole own area and a lot harder. But in my opinion, machine learning provodes tools to do so which I will try to show further on in this blog!

      The basic idea is the same as used when making an Xg – plot of a past game. Training data – data from earlier events/games to build the model on and then use that model to predict future events/Xg or games…


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s