The only slightly more advanced than shots/possession statistics thread

[Hound]

Nice one, thanks

[Brain Traysers]

One of the more advanced player-based metrics slowly dripping down the leagues to the EFL (where there is generally worse quality and quantity data availability) is Expected Threat (xT).

It aims to solve one of the criticisms of xG, namely that there are many other steps before a shot takes place. xT attempts to quantify the value of moving the ball from one area to another (thereby increasing the goal "threat"). The xT scores can be best understood as "probability that the team score from possession in the current area in the next 5 moves". Having the ball in the centre circle is an xT of around 1%, on the D around 10% and around the penalty spot 17%. The metric was introduced a couple of years ago by Karun Singh (who was subsequently hired by Arsenal), and the full write up (inc v good tools to play with) is available here.

Here is a plot of xT (per 100 possessions) compared with frequency of actions (how involved a player is in possession) for midfielders last season.

- Unsurprisingly, our stand out player is Swift (0.65 xT per 100 actions). Most of the players with higher xT per 100 actions have fewer overall actions (so potential small sample issues). It's striking how busy he is - only a handful of players averaged more than his 46 Total actions p90, and only 2 of those managed higher xT per 100 actions
- Rinomhota (c. 0.35 & 31) and Pele (c. 0.44 & 40) appear to be the other RFC players fitting the "midfielder" filter - both are below avg for xT, and involved in an average amount of possessions. I can't see Ovie, so he may not be classified as a midfielder, but I see Eze (or rather, his minutes explicitly playing as a "midfielder", as his dot is small) at c. 0.42 & 37, 'busy but below avg threat' which is interesting.
- Away from RFC, Pablo Hernandez is far too good for this league, while Tom Cairney's game has evolved to be very busy but not involved much in an attacking sense which seems a fair reflection of the view that Fulham got promoted in spite of Parker, rather than because of him.

Of course, as with any model this is a simplification - in this case the biggest issue is the lack of player tracking data. In xT having the ball on the edge of the box with teammates around and no defenders in sight, is treated as the same as having the ball in the same spot with 11 opposition players ahead of you and no team mates around. But nevertheless, it's another step forward in the public analytics arena.