Things just keep moving.
A decade ago, possession percentage wasn’t yet on peoples’ minds; now, the biggest symbol of stats in football — expected goals (xG) — is calculated with computer modelling. That’s quite a change.
However, just because there’s an algorithm involved doesn’t mean it’s difficult to understand. The concept is pretty intuitive, even if the language around it isn’t always so clear.
We’ve decided to write about five big metrics (including expected goals), what they mean, and what they can be used for. Keep an eye out for them in 2021.
1. Expected goals
Expected goals is a way of judging how good an opportunity was, essentially a snapshot at a moment in time, right as a player takes the shot.
The models that calculate xG (like ours, for example) take into account things like where the shot was taken and the passage of play leading up to it. The models use hundreds of thousands of shots, and because they know which ones turned into goals they can see what kind of influence each of these things has.
The statistic is useful for knowing how good teams are at creating chances, and how good forwards are at getting opportunities. This latter point is an important one. The general analytics consensus, using work with expected goals, is that the gap between a player’s ability to get chances — which can be gauged with xG — is a more significant influence on what makes a good striker than the player’s ‘finishing ability’.
2. Post-shot expected goals
While expected goals takes a snapshot at the moment a player hits the ball, post-shot expected goals adds the general direction the shot is taking into the mix. At its simplest, if something is going towards the top corner it’ll be worth more than if it goes low down the centre of the goal. The reference to ‘post-shot’ is because we only know this information after the shot has been struck.
Post-shot expected goals can be a useful descriptive stepping stone between expected goals (occasionally referred to as ‘pre-shot’ to differentiate it from post-shot xG) and actual goals.
For example, if a player has taken shots worth 20.0 expected goals over the course of a season but only scored 15, there may be several reasons for this. If their post-shot expected goals figure was 15.5, then it’s likely that they’ve struggled with their finishing (either missing the target completely or making their attempts easy for goalkeepers to save).
Where post-shot expected goals has a stronger analytical value is for goalkeepers. Most post-shot expected goals models give blocked and off-target shots a value of 0. Due to this, post-shot xG represents the value of the shots that goalkeepers actually face. This can be compared to the number of actual goals conceded, just like pre-shot xG can be compared with the number of goals scored for strikers.
The above shot map – using post-shot expected goals – shows that Kepa Arrizabalaga has conceded all of the close, central shots he’s faced in the 2020/21 Premier League, but that these were generally difficult shots to stop.
3. Expected goals assisted
Expected goals assisted (sometimes shortened to xA or xGA) is a simple concept. The player who sets up a shot gets credited with the xG value of that shot. If a striker takes a shot worth 0.33 expected goals, the player who set it up will get 0.33 expected goals assisted added to their account.
Just like expected goals can be a useful stat for judging forwards, expected goals assisted can be useful in judging creators. Not all shots are the same value, which means that not all key passes are.
PPDA stands for ‘opponent passes per defensive action’. Like MPH for ‘miles per hour’, the name is a simple acronym to indicate how many passes an opponent makes for every defensive action that a team makes. A lower PPDA figure indicates a more intense pressing style.
PPDA is generally limited to a certain area of the pitch. The original development of the stat — by Colin Trainor in 2014 after discussions with Rene Maric — discounted the deepest 40% of the pitch for the defensive side. The reasoning for this is that the metric seeks to look at a team’s pressing, and when you get too close to your own goal this isn’t really ‘pressing’ any longer.
Technically, then, PPDA is ‘opponent passes per defensive action in the furthest forward three-fifths of the pitch from the defending team’s perspective’, but that’s not as snappy an acronym.
Possessions and sequences are more building blocks than widely-used statistics in their own right.
Both are spells of one team having possession of the ball, the difference is that possessions can be made up of multiple sequences. For example, if a team has the ball, takes a shot, and gets a corner that would be two sequences (the original move and then the corner) but one single possession (as the team never truly lost possession of the ball).
These two metrics can be used on their own to provide useful and/or interesting information. The number of sequences in a match can be an indication of how frenetic it is, and teams who usually have these high-sequence-count games will usually have a different style of play to teams who have low-sequence-count games.
However, possessions and sequences become more useful when combined with other things. ‘Sequences that feature ten or more passes’ is a nice stat to indicate teams who like to keep the ball a lot. There may be overlap with a general possession share percentage, but ‘sequences of 10+ passes’ can be a more tangible and specific metric.
If you’d like to learn more about our products or services, and how they might be able to help you, don’t hesitate to get in touch.