What is it?
Chances created = assists + key passes.
Assists = the final pass leading to the recipient of the ball scoring a goal.
Key passes = the final pass leading to the recipient of the ball having an attempt at goal without scoring.
Why Bother With Chances Created?
Goals, shots on target, and possession each paint a corner of the big picture (e.g. the effectiveness of a team’s attack).
For example, if you only focused on goals, you’d be underrating how efficient Burnley is since they’re 19th in goals scored (as of 16th Feb 2018). If you only focused on possession %, you might be surprised to see Swansea sitting in 10th in that category, and they are actually 16th in the league table.
Ideally, we want a measure of a team’s attack that correlates with their overall success. Fortunately, research conducted that shows “chances created” has a high correlation with table position. This paper reinforces such a view.
Drawbacks And Solutions
Here’s a rundown of issues with chances created, and what can be done to account for them:
One limitation that is avoidable is that it only focuses on events in the oppositions final third. No solution for this yet.
Shots Leading to Created Chances
Chances created is currently limited to the “final pass” leading to a shot attempt. It should also account for rebounds off of shot attempts, like Gaye’s 1st goal against Bournemouth.
This is fairly easy to solve. If key pass numbers are added to shot attempts, a fuller attacking picture is manifested. And since some players don’t play the full 90 mins, adjusted playing time should be accounted for.
So if CC = chances created plus shots, KP90 = key passes per 90 mins, and S = shots per 90 mins, our formula looks like this right now:
The Variance of Chance Quality
Much like how there are different probabilities for every goal attempt, not all chances created are equally valuable. Being able to take into account the success rate would give a better indication of quality.
If we insert two metrics here, Assist Rate (“AR”) and Scoring Rate (“SR”), we can reward players for being efficient and effective with their play. Assist Rate can simply be Assists/Key Passes, while Scoring Rate can be Goals/Shots. Adding that into the equation now looks like this:
Different Positions, Similar Expectations
Should Lascelles’ chances created output be compared to Perez? Or Diame’s output compared to Gayle? Limiting comparisons of a plyer’s output against the Premier League’s average to positional peers could give better context for how relatively good or bad a performance is.
If we take the CC+ output and compare to the CC+ average for a given player’s position, the new equation looks like this:
With this equation, we should be able to see just how creative (or uninventive) a player is compared to others at their position.
Chances Created In A Match
To test out these equations, let’s look at the 11 Feb 2018 NUFC v MUFC match, here’s a comparison of chances created for each side:
Now let’s see how it looks with the “weighted chances created” (or wCC+) equation:
Suddenly, we’re able to see the larger picture regarding who was productive, or unproductive. For Man Utd, Martial went from only having 1 chance created to looking as if he were a major contributor to Man U’s attack.
One thing we can do to help demonstrate how “good” someone was for given position is to normalize these numbers to 100. If we make it so that 100 is bang-on average, it’ll be easier to see respective deviations.
Here are the weighted chances created for the NEW-MNU game, normalized to 100 (100=average):
Again, we have fresh analysis points. Shelvey was far and away Newcastle’s most creative player (relative to his position), and his rating here is actually pretty close to his season average (as we’ll see shortly.)
For Man United, Pogba’s bad game looks even worse in this context. He came up utterly empty-handed in 65 minutes, while his direct replace Carrick was decently productive.
Now let’s apply these equations to the Premier League clubs, its players, and the Newcastle squad, and see what we find.
Newcastle vs. the PL, in (weighted) Chances Created
Each PL Clubs Average wCC+ Output, per Match:
Note: this data was generated after Week 27, where Newcastle sat 13th in the table.
From this data output, one can start conjuring up a general description of a club’s attack.
- Man City is 1st in Goals Scored, Possession %, and wCC. This suggests they are clinical and can create a high volume of chances from a high share of possession. (smell test = passed.)
- Swansea is tied for 20th in Goals Scored, 14th in Possession %, and 20th in wCC. This suggests that despite being somewhat effective at retaining the ball, they aren’t clinical and they produce a low volume of chances.
- Newcastle is 14th in Goals Scored, 19th in Possession %, and 11th in wCC. This suggests that they are below average in converting shots and are poor at retaining the ball, yet are effective at creating chances.
Player Progression Over A Season
Another way that the wCC+ metric can be applied is to observe how a player performs over a season. Because the metric pulls more data points, there are more opportunities for trends to form.
Mo Diame, as an example, is a player that had poor showings in the early part of the ‘17-’18 season and seemed to have an uptick in form as the season progressed. To see if the numbers correlate with this sentiment, here are his game-by-game wCC+ returns:
While there have definitely been more consistent contributions in the 2nd half of the season (spiked with great isolated performances), on the whole Diame hasn’t been a dynamite offensive spark.
To clarify, this doesn’t mean he hasn’t added value elsewhere on the pitch, as this only looks at his ability to create scoring chances.
Chances Created Rankings (by Position Groups)
We didn’t want to compare players that are regular starters against the odd man that’s played 20 minutes all season (getting rid of the pesky small samples), so we eliminated players that appeared in less than half of games played (as of 17th February 2018).
Here are the positional group rankings of wCC+ (Newcastle players are highlighted).
PL Comparison Notes:
- The NUFC midfield’s ability to create chances is quite mixed – Shelvey is a standout, Merino is near-average, and Diame & Hayden are very poor
- All our attacking mids drag down the average, with Ritchie being comparably average for his positions expected output
- The NUFC centerbacks rate as being slightly above average in creating chances, which might be an indicator of NUFC’s counter-attacking prowess
- NUFC full backs are nearly non-existent in reliably creating chances
- Gayle & Joselu are definitely below-average in creative contributions, though they are closer to the middle
The NUFC Squad’s wCC+
Here are the wCC+ numbers for the full Newcastle squad (as of 17th Feb 2018):
Going purely off of this, it looks as if Matt Ritchie is our “best” offensive contributor. Now here’s each respective position normalized to a 100.
Normalized NUFC wCC+ Rankings
NUFC Squad wCC+ observations:
- Our center backs are the most productive, which perhaps can be linked to our tendency to launch long balls for counter-attacks
- Our full backs are not creatively influential. At all.
- Gayle & Joselu’s numbers are nearly identical, which suggests that the striker output is more systemic than it is individual performance
- Shelvey’s productivity is remarkable, considering how separated from the rest of the central mids he is
Bear in mind, this is strictly an analysis of a player’s ability to create chances.
There are other aspects of attacking, and innumerable facets to overall match performance. For example, while Jonjo Shelvey is far and away Newcastle’s most reliable creator, he is the worst contributor to defensive actions in the midfield.
Having said that, nearly everyone is creating below-average for their position, while there are a few (i.e. Ritchie & Merino) that are landing at roughly average. It’s clear that the squad’s attacking quality overall has ways to go, in terms of catching up to the Top 10.