Does Passing Accuracy Predict Results?

photo courtesy Joffley/photopin cc

By Bill Farnham

Is there a relationship between Passing Accuracy and Passing Forward and Liverpool’s results in the Premier League this season?
(and can player’s stats demonstrate this relationship in a plausible way?)

If I tell you that Glen Johnson, Joe Allen, and Steven Gerrard are important players for Liverpool and that their performance in a match has an impact on the outcome of that match, is it ok then to say that their performances (positive or negative) should correlate in some way to the result achieved in the match? When they (the players) can be said to have played well, does Liverpool also do well?

Try to read this part now before you go too much further.  It will make life much easier (for me, mostly):

  1. These stats are for Premier League Matches only and come from EPL Index
  2. I chose Johnson, Allen, and Gerrard because they have played in the majority of Liverpool’s matches in the Premier League this season and because they are acknowledged to be important players for the team.
  3. A positive correlation between an outcome and a statistic means that the outcome is more likely to occur, while a negative correlation would similarly indicate that the outcome is less likely to occur. Obviously, the higher the correlation percentage, the greater the potential for the outcome to occur.
  4. Correlation does not imply causality according to wikipedia (“The Pearson correlation coefficient indicates the strength of a linear relationship between two variables, but its value generally does not completely characterize their relationship.”), however a correlation can indicate that there might be a causal relationship.

Now, back to where we were…

Extending this logic one step further, over a series of matches then there should be a correlation between specific performance attributes of important players and the outcomes of those matches, if the performance attribute in question has a relationship (is meaningful) to the result.

For example, there should be a correlation between Liverpool’s passing accuracy in a match and the number of points Liverpool takes from a match if there is a relationship between Passing Accuracy percentage and winning a match. The more I think about it, the more I would probably expect to see a positive correlation between Passing Accuracy and Points, i.e. the greater the Passing Accuracy the greater the number of points taken from a match.

Using Johnson, Allen, and Gerrard as data points I’m going to see what the stats say (insert usual disclaimers, caveats, yada yada, drivel, etc.).

The two performance attributes I evaluate are:

    1. Passing Accuracy (PA percentage), defined as the number of completed passes divided by the number of passes attempted.  For passing accuracy, the direction of the pass does not matter.
    2. Passing Forward (PF percentage), defined as the number of forward passes divided by the number of passes attempted.  Forward Passing attempts can be lower the higher up the pitch a player plays; defenders will often have a higher forward passing percentage than advanced midfielders and strikers, for example, simply as a result of where they line up on the field.

Here’s how I would expect the performance attributes to correlate with results:

    1. PA Percentage: I believe PA percentage should correlate positively with results for the team as a whole, as the greater the passing accuracy for the team, the better the chance that the team is retaining possession.  Dominating possession increases a team’s chances of success in any given match, as it is impossible for the opposition to score when they do not have the ball.  (Michael Cox of Zonal Marking writes a very interesting analysis for the Guardian here)
    2. PF Percentage: I don’t have any preconceived notions of how PF percentage should correlate with results for the team as a whole, however on a player by player basis, I would expect PF percentage to correlate positively with match results for players viewed as important offensive contributors, since good players passing towards the opponent’s goal can only be a positive thing as far as wins are concerned.

Let’s look at what the statistics say for the first 16 matches of the season (Note: Glenn Johnson’s stats are only for 14 games, as he missed match rounds 9 and 10).

In this graph, the correlation between points achieved and PA percentage for Johnson, Allen, and Gerrard is displayed. While Allen has a very slight positive correlation with points achieved and Gerrard has a very slight negative correlation (the more accurate his passing has been this season, the less likely Liverpool has been to get all three points), the interesting data point is the moderate negative correlation between Johnson’s PA percentage and points earned. These numbers indicate that as Johnson’s passing accuracy increases, Liverpool’s chances of realizing three points from a match decreases.

Here are Johnson’s numbers sorted by decreasing PA percentage:

Player Match Round PA% PF% Points
Glenn Johnson 4 92% 15% 1
Glenn Johnson 1 91% 26% 0
Glenn Johnson 12 87% 43% 3
Glenn Johnson 5 86% 29% 0
Glenn Johnson 13 86% 33% 1
Glenn Johnson 3 84% 34% 0
Glenn Johnson 11 84% 37% 1
Glenn Johnson 6 83% 40% 3
Glenn Johnson 14 83% 30% 0
Glenn Johnson 16 83% 42% 3
Glenn Johnson 2 81% 52% 1
Glenn Johnson 15 80% 38% 3
Glenn Johnson 7 79% 38% 1
Glenn Johnson 8 76% 36% 3

If anyone reading this has any interesting ideas that could explain, please comment.  My interpretation is that in situations where Johnson is not moving forward and pushing higher up the pitch, i.e. in situations where his passing is more likely to be short and less adventuresome, Liverpool is less likely to win.  Put another way, if Johnson is pushed back by an attacking winger and opposition fullback pairing, his passing tends to be more sideways and backwards and low risk, not the type of passing that has a great impact on goal scoring, and subsequently Liverpool are not attacking frequently in the match.

In the graph above, the PF percentage from both Johnson and Allen demonstrate interesting correlation percentages with points earned in a match. Perhaps the most interesting data point is that Allen’s PF percentage has a small negative correlation with points earned.

Here are Allen’s numbers sorted by decreasing PF percentage:

Player Match Round PA% PF% Points
Joe Allen 5 88% 40% 0
Joe Allen 13 93% 38% 1
Joe Allen 9 86% 37% 1
Joe Allen 7 91% 35% 1
Joe Allen 16 87% 34% 3
Joe Allen 12 93% 33% 3
Joe Allen 14 87% 33% 0
Joe Allen 1 96% 32% 0
Joe Allen 3 93% 31% 0
Joe Allen 6 96% 29% 3
Joe Allen 2 93% 28% 1
Joe Allen 11 89% 28% 1
Joe Allen 4 91% 26% 1
Joe Allen 8 95% 25% 3

This result says to me that maybe in matches where Allen has a greater responsibility for bringing the ball forward from the midfield and passing forward, Liverpool is less likely to earn three points from a match.

The counterpoint to Joe Allen, however, is Glenn Johnson, whose PF percentage correlates strongly with Liverpool realizing points in a match. This information also seems to corroborate the correlation seen in the graph of PA percentage. As Glen Johnson is making more forward passes, passes more likely to be defended against, his PA percentage will decrease, but Liverpool is more likely to triumph.

Courtey photopin cc toksuede

Final note: Steven Gerrard does not stand out statistically in either of these categories, which is a bit surprising, perhaps, given the conventional wisdom regarding his importance to Liverpool. I’m choosing to believe that other statistical measures will be more relevant in sussing out his impact to the team and its chances of winning. Next week I’ll analyze two additional statistics, Total Chances Created and Penalty Area Entries.


2 thoughts on “Does Passing Accuracy Predict Results?

  1. There needs to be more quantitative analysis applied to soccer so that we can get beyond the pure subjective statements. Thanks for getting at it — and starting to ‘get at’ what the correlation is between what some of the key players are doing in a game and whether that means the team is doing ‘winning or losing’ the battle on the pitch.


  2. Pingback: Does Passing Accuracy Predict Results, part II | The Red Letter

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s