Karun Singh Profile picture
16 Feb, 17 tweets, 11 min read
2 years since Expected Threat (xT) was first published! Thought I’d highlight some of the exciting follow-up work that various people have done since:
First, a quick statistic that I had a hard time believing – 26,000+ unique people have visited the post and logged a total of 120.8 days (i.e., 4 months) of reading time! Image
[Explainers]

If reading a blog post isn’t your cup of tea, this recent podcast by @FanalyticsPod (@ryanbailey57 and @MarkCarey93) is an excellent introduction to xT:

[Explainers]

If you prefer watching a talk, here’s one from the StatsBomb conference (Oct ‘19) that describes the intuition behind xT and its various applications:



(slides: tiny.cc/xt-statsbomb)
[Extensions & Applications]

Continuing with the StatsBomb conference, @robwhickman explains how he extended xT to take defensive risk into account:

[Extensions & Applications]

@rjtavares / @lastrowview describes a method of evaluating off-the-ball player movements by combining xT and tracking data:

medium.com/football-crunc…
[Extensions & Applications]

@dlareg49 proposed Joint Expected Threat (JxT) as a way of quantifying player chemistry, and incorporates this metric into various visualizations:

[Extensions & Applications]

@dperdomomeza1, @ddgirela, @EveryTeam_Mark, and James Goldring (Twenty3 Sport) used the xT framework, among many other techniques, to study how one can attack set defences:

[Extensions & Applications]

In this piece for ASA, @arjun_balaraman uses xT to identify MLS strikers that excel at creating shooting opportunities for themselves (in contrast to pure goal poachers):

[Extensions & Applications]

Since xT doesn’t make any sport-specific assumptions, it can be adapted to other settings fairly easily. One example is @SamForstner and @yuorme's work on xT for ice hockey:

[Extensions & Applications]

@rwohan integrated xT very effectively into a couple of his data-driven pieces to explain Liverpool’s anomalous midfield and Arteta’s cryptic comments:

espn.com/soccer/english…

nograssintheclouds.substack.com/p/what-the-hec…
[Visualizations]

xT lends an additional level of granularity to xG-based visualizations. For example, we can now render game momentum timelines in addition to xG timelines, and danger heatmaps in addition to shot maps:

ImageImage
[Visualizations]

@markrstats is one to follow if you’re interested in more xT-based visualizations; a couple of recent examples:



[Comparisons]

@MaaikeVanRoy, @p_robberechts, @TomDecroos, and @jessejdavis1 comprehensively compared xT and VAEP along several dimensions and highlighted the fundamental tradeoffs each framework makes:

dtai.cs.kuleuven.be/sports/blog/va…
[Comparisons]

Their group also occasionally publishes data-driven season reviews that contrast the players and actions highly valued by VAEP and xT:

[Comparisons]

How does xT stack up against xG as a predictive signal? @MilanKlaasman sheds more light on this topic here:

- Finally, if you’d like to use xT values off-the-shelf, this is a good place to start:

- If you’d like to build your own xT model or extend it, feel free to reach out and I can provide pointers!

If I missed something, please let me know!

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Karun Singh

Karun Singh Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!