Events in football are nigh-impossible to quantify, the defensive side of things doubly so. Whereas an attacker’s “good day” generally involves easily measurable events (goals, shot attempts), a quality defensive performance should be a quiet, stat-barren one.

Measuring the quality of a side’s defense, let alone individual defensive performances can be a fool’s errand. However, if defensive responsibility is viewed as variations of ‘defensive pressure’, a workaround becomes possible.

Note: defensive pressure is different from the concept of pressing – everyone applies defensive pressure at some point, but not every presses.  Here are the 3 main areas in which defensive pressure is applied to the opposition:

PPDA

A way that defensive pressure can be quantified is through a metric called “Passes Per Defensive Action”. The concept is this: take the number of passes the opposition attempted, and divide that by the sum of the 4 proactive defensive stats (interceptions, successful tackles, failed tackles, and fouls).

A low number (e.g. 5.0 PPDA) means that the opposition was only connecting 5 passes for every defensive action. A high number (e.g. 25.0 PPDA) means that over 17 passes were typically attempted before a proactive challenge was made.

What’s useful about it?

Every team has its own unique approach to defending, and PPDA doesn’t favor sides that defend in low, mid, or high blocks.

In fact, without having to pore over hours of match film, looking at a club’s PPDA average can give a sense of how much defensive pressure they typically employ.

Analyzed over time, it can also be a descriptor for a club’s performance. If Newcastle typically have a 11.0 PPDA average and allow 29 PPDA in a game, that would indicate that they had little influence on the balance of play. Similarly, if they had 5 PPDA on a given day, that’d indicate that had an unusually aggressive approach.

What’s misleading about it?

Since there’s no information about shot attempts, PPDA is confined to being a possession-related stat. Possession can be an indicator of match influence, but it doesn’t correlate to points won. In fact, there is little correlation between high PPDA numbers and Points Won.

To illustrate, here’s a scatter plot of each PL club’s PPDA average against the points they’ve won, through Week 15 of the ‘17-’18 season.

Notice how Liverpool & Southampton have a near-identical PPDA average and yet their season form couldn’t be more different. Another comparison to consider is Burnley & West Brom – WBA actually average a lower PPDA output, and they are currently 11 clubs between these two in the table.

To put it simply – PPDA is more a descriptor of a team’s defensive approach than it is a predictor of match outcomes.

NUFC’s Season vs. PPDA

From the Expected Goals piece, the stat xGD was found to be an alternative to discerning a team’s form over the season. By using xGD as a baseline, Newcastle’s PPDA could potentially show some correlation — for example, if they had an unusually negative xGD rating in a match, one could assume that they had little influence over the balance of play (which would imply a high PPDA output).

Here are the results of that hypothesis. The x-axis is xGD (the higher the better), the y-axis is the PPDA rating (the lower it is, the higher the defensive pressure), and the dot size represents points won from each game (0=small, 1= slightly bigger, 3=biggest):

It turns out there isn’t a ton of correlation between xGD & PPDA. (The p-value & r-squared values are pretty low.) However, there are a couple of takeaways here:

  • Tottenham in Week 1 – having Shelvey sent off didn’t help matters. Who knew?
  • Our worst performances were against Man City (home and away), Liverpool (home and away), and Manchester United (away) and Arsenal (away) – nothing to see here.
  • The closer Newcastle’s PPDA is to its season average (11), the more likely points are won.
  • If a game’s PPDA > 11, the odds of points won greatly increase.
  • That Week 35 West Brom game is quite the statistical outlier.

PL Position Rankings, by Defensive Actions

Trying to determine the PPDA for individuals hurts my brain just thinking about it. Not only would every defensive zone have to be mapped across opposition passes .. no, I’m just stopping there.

However, if the “defensive actions” part of the concept is used, the defensive productivity of individuals can be found. We can measure both how often a defensive action is performed (the rate of challenges), and how successful they tend to be (the rate of success).

Like all stats studies, context is important: how a striker performs in defense is different than how a defender does. So all outfield players were separated into 3 main groups: forwards (the first line of defense), midfielders (the primary source of defensive pressure), and defenders (the last line of defense).

As always, in order to keep outliers from mucking up the data, only players with over 15 appearances were used (Kenedy being an exception):

 

Forwards

  • Ayoze Pérez comes in with as one of the few best amongst all 9s & 10s. Is one of the most aggressive and most successful in his challenges.
  • Joselu comes out looking decent here. A fairly high challenge rate, and as close as it gets to an average successful:failed challenge rate.
  • Gayle is a bit below-average in challenge rate and success rate.

Midfielders

  • Much like Pérez, Diamé is in the top percentiles in his position group. Only Wilfred Ndidi is more aggressive than Mo, and Diamé is consistently successful in his challenges.
  • Merino & Kenedy occupy the same space – above average in challenge rate, and about typical for the position in success rate.
  • Ritchie & Hayden also function similarly – below-average in challenge rate, and also about average for success rate.
  • Jonjo Shelvey is also below-average in how often he’ll perform a defensive action, but he’s our best midfielder in terms of success rate. (I personally did not see that coming.)
  • Jacob Murphy & Christian Atsu both pull down the average in both challenge & success rates. Essentially, they are defensive voids on the pitch.

Defenders

  • Paul Dummett is an assured defender. While the rate of his challenges is a bit below-average, there are only 10 defenders in the entire top-flight that have a better ratio of success:fail defensive actions.
  • Lascelles having a “low challenge rate” (read: Jamaal Lascelles) isn’t necessarily bad – remember, great defensive often won’t show up on the stat sheet. And When he does get called into action, he does have a strong ratio of successful defensive actions.
  • Lejeune’s success rate is nailed on “average” for defenders, and the rate of his challenges is slightly higher than normal.
  • This study does Ciaran Clark little favor. His challenge and success rates are among the worst in this whole group.
  • DeAndre Yedlin comes out ok. While his challenge rate is high, it is negated by a poor success rate.
  • Manquillo has a slightly above-average challenge rate and, like Clark, has one of the worst success rates among all defenders.
  • Based off of this study alone, it should be noted that Lejeune & Lascelles make for a more complementary pair rather than Clark & Lascelles. It can also be inferred that Clark & Lejeune could mesh well (in reality though, results have been mixed).

NUFC Squad Comparison of Defensive Actions

If Newcastle’s squad is thrown into the same group, and their ratings are normalized to their respective positions (100 = position average), the output looks like this:

  • On one hand, Murphy’s extremely low challenge & success rate needs to be cut some slack – it’s fair to say that these will be much better after more time under Rafa & co. However, Kenedy is leagues ahead and is a year younger than Murphy.
  • Speaking of – look at Jonjo Shelvey! Might not challenge often, but when he does it has a high success rate.
  • Mikel Merino is without a doubt the most proactive player we have. While this graph doesn’t show it, he has an unfortunate knack for fouling in his challenges – if that gets sorted, he will probably own this realm in the future.

Conclusion

From an overall team standpoint, PPDA can help show how healthy defensive pressure was for a given game. The lower Newcastle’s PPDA output has been, the more likely points are won. PPDA can also inform how aggressive the squad’s defensive pressure was (or wasn’t).

From an individual standpoint, defensive actions show that while there are some gems in the squad, our fullbacks and wingers aren’t particularly proficient at applying defensive pressure. There is room for higher quality on the defensive side of things.