[PyMCon Web Series 08] A Soccer-Factor-Model to Decipher a Soccer-Player's Inherent Skill (July 17, 2023) (Maximilian Goebel)

Welcome to the 8th event of the PyMCon Web Series!

Speaker: Maximilian Goebel, a Post-Doc in Finance/Economics at Bocconi University
Event type: Live Talk
Date: 2023-07-17T14:00:00Z (subscribe here for email updates)
Time: 2 pm UTC / 7 am PT / 10 am ET / 4 pm Berlin
Register for the event: Meetup event (to get the Zoom link)
Website: PyMCon Events · PyMCon Web Series

Abstract of the Talk:

Inspired by the asset-pricing literature, the Soccer-Factor-Model (SFM) is an attempt to determine a soccer player’s “alpha”, i.e. his/her inherent skill. In this application, “skill” is defined as a player’s probability to score a goal after accounting for factors that are actually to be attributed to his team’s out- or under-performance vis-à-vis the opponent. In that sense, the SFM tries to answer the question, whether a player’s observed goals are an over- or understatement of his “true” skill/ability.


Max’s Interview:

Live Talk:


About the Speaker:

Maximilian Goebel

Max is a Post-Doc at Università Bocconi, specializing in finance/economics. With research interests in climate econometrics, machine learning, macroeconomic forecasting, and asset-pricing, Max combines expertise in these areas to drive innovation and uncover valuable insights.

Connect with Max:


LinkedIn: https://www.linkedin.com/in/maximilian-göbel-188b0413a/


We thank our sponsors for supporting PyMC and the PyMCon Web Series. If you would like to sponsor us, contact us for more information.

As the place where the future of AI gathers, ODSC works to bring together the global data science community to encourage the exchange of innovative ideas and the growth of open source software." or as well you can use the short sentences like “The future of AI gathers here”, “Build AI better”.

The upcoming events include ODSC Europe (June 14-15, in-person (London) and virtual), Generative AI Summit (July 20, virtual), ODSC APAC (Aug 22-23, virtual), ODSC West (Oct 31-Nov 3, in-person (San Francisco) and virtual).

Connecting with PyMC


FYI @RavinKumar, in case you see this during the talk, the Zoom chat is disabled.

edit: no longer disabled after the link was sent at 10:04.


The motivating idea behind your features basically seems to be establishing the goal scoring with/without a given player - effectively using the time with them not on the field as a “control” and with them on the field as a “treatment”. If that’s correct, does this implicitly assumes a defensive team will respond the same way agnostic of who is on the field? Is there a good way to consider a change in the defensive behavior when a certain player is on the field? I could see this being the case for a superstar, for example. Or do you think that information is contained in the scoring difference?

1 Like

How would you control for skill over time? Basically, estimating the “aging curve” of a player?


do the features and included rows account for when players are on the team but are not playing? For instance, a player on the bench would influence the probability of the team scoring

1 Like

Thanks for the response! Even the diff-in-diff idea is a good thought!

1 Like

seems related to the questions above with “superstars”

In case you missed the live session, the video has been posted to YouTube.

Watch the recording here:

1 Like