📊What are the characteristics of successful Premier League teams?
Investigating how strongly a range of team metrics correlate with points accumulated in a season, using data from the last 5 Prem seasons.
Correlation scores indicate strength & direction of correlation (1/5)
Correlation scores are in the range [-1, 1], with magnitude > 0.4 indicating moderate correlation with points.
We know that goal diff and points will be closely related, and we include it to demonstrate a metric that correlates extremely strongly with points: r =0.97 (2/5)
On the other hand, we see some metrics that appear to have absolutely no correlation with points accumulated.
Percentage of expected threat (xT) conceded by opposition crosses is a great example of this: r = -0.02 (3/5)
Plenty more in-depth work to come on this, including a discussion/digestion of the initial results shown.
As a quick preview, we can look at the linear correlation between any pair of metrics using a correlation matrix. Please zoom in!🔎 (4/5)
A few notes: (i) All metrics have been generated by aggregating Opta event data over the past 5 Prem seasons. (ii) Some metrics are normalised by ball losses as a proxy for "per possession". (iii) As always, code is available on GitHub:
(5/5)github.com/jakeyk11/footb…
Brief definitions of metrics used:
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.