Defensive action, pressure, and passing

A preliminary attempt to identify two types of players based on defensive stats in the 2016 NWSL season.

Kim Little defending against Christen Press or Christen Press defending against Kim Little? // Credit: Nikita Taparia

Last week, I illustrated a fairly simple trend on passes under pressure and consistency on Backline Soccer. This week, I would like to bring defense back into the equation because passes under pressure implies the opposing team player is creating this pressure. First, I will review some terminology for defensive action and pressure with updated graphics (78 games). I previously talked about it defensive play on the blog. Then, I will examine passing and defense for each team and position. I will preface by stating — most of this seems very obvious but the differences between teams lend a bit more insight.

Figure 1. Defensive action can be divided between ball and player defending. Defenders are more conservative in defense. Forwards do not defend as much (as expected). Midfielders display the largest differences with Portland being the most aggressive and Western New York Flash being the most conservative. (78 games, at least 270 minutes for players)

Some time ago, I introduced the notion of conservative and aggressive defensive action. Conservative actions include ball shields, blocks, clearances, and interceptions. Aggressive action includes tackles, dispossession, and challenges. The division between the two actions illustrated that defenders (blue) took more conservative defensive actions, midfielders (green) had the largest range of differences, and forwards (orange) were clustered to have very little defensive action (Figure 1). Specifically for the midfielders, Portland, Western New York, and Chicago had more defensive actions per game and Houston, Seattle and Kansas City logged the least number of actions.

Figure 2. A player will either pressure their opponent to make a mistake or take physical action to retrieve the ball and gain possession. Here, you can see how defenders and forwards form clusters and differ in defensive action while midfielders have the most differences depending on team. (78 games, at least 270 minutes for players)

When this was first introduced, I separated defensive pressure from defensive action and argued a player will pressure their opponent into making a mistake rather than taking action. When action and pressure per game are compared to each other (Figure 2), forwards and defenders form clusters that differ in total defensive actions per game. Midfielders have the largest discrepancies between teams, in which Sky Blue applies the most defensive pressure and Portland and Chicago midfielders have the highest number all around for pressure and actions per game.

There are a few things that came to mind when I first did this exploration. More defensive actions can mean either they are mostly on defense or they just take more initiative. If a player is not performing a defensive action or applying pressure, it is because they have possession and must pass under pressure. To understand this, we can create a ratio between defensive pressure and action per game. For example, a ratio of 4.0 means a player applied four times more defensive pressure than takes action. A ratio of 0.5 means a player takes double the defensive action than pressure.

Figure 3. On a scale for pressure to action, forwards apply the most pressure and pass under the most pressure. Defenders take the most action and receive the least amount of pressure. Midfielders define a larger range. (78 games, at least 270 minutes for players)

There are a few confirmations with a graphic (Figure 3) that compares defense to passes under pressure. Forwards apply more defensive pressure than action and they must pass under pressure more than anyone else. This makes sense, defenders are on their tail and they are constantly trying to force them to make a mistake so they can repossess the ball. Defenders take more defensive action than pressure and pass under less pressure. This also makes sense because often times, defenders set the play up again and want to cleanly take the ball out of play without giving up a foul. As always, the largest range of differences lies with the midfielders. The midfielders represent the transition from defense to attack. If I were to hypothesize, we can begin to dissect which midfielders act more as defenders or forwards but hold this thought for just a moment!

We can break defensive action apart in a similar fashion to pressure and action. Here, the ratio of conservative defensive actions to aggressive defensive actions illustrates a type of action taken by a player on a particular team and position.Thus, a ratio of 2.0 indicates double the conservative action compared to aggressive action and a ratio of 0.5 indicates double the aggressive action compared to conservative action.

Figure 4. On a scale for conservative to aggressive, forwards are more aggressive and must pass under the most pressure. Defenders are more conservative and pass under the least pressure. Midfielders exhibit a range of either conservative or aggressive. Defenders also have a spread. (78 games, at least 270 minutes for players)

With this graphic (Figure 4), we can confirm of a few obvious statements. First, defenders take more conservative actions and pass under the least amount of pressure. This makes sense since aggressive actions are riskier and could lead to a penalty or dangerous free kick. In this case, you can actually see a bit more spread in defensive types — Seattle with the most conservative and Western New York Flash with the most aggressive. Second, forwards take more aggressive actions. This also make sense because forwards want to switch the point of attack and can afford to take more risks but they also must face these conservative defenders and pass under higher pressure. Finally, as always, midfielders have a huge range of differences and in this case, the divide between a defensive midfield and an attacking midfield is much clearer.

Figure 5. Cluster analysis reveals two major clusters (circles and triangles). To read this graph, use the example point, which denotes a player that applies defensive pressure 3 times more than action and of the action taken, conservative is 2 times higher than aggressive. (78 games, at least 270 minutes for players)

As a first attempt to differentiate between attack and defense, cluster analysis can take into account the parameters discussed throughout this article and group similar points. The graphic compares the two different ratios discussed earlier (Figure 5). Percentage of passes made under pressure is denoted by the size of the data point. Two clusters were found (circles and triangles). From this, the midfielders are grouped either with the defenders or the forwards. Specifically, Western New York, Houston, Chicago, Orlando, and Portland are grouped with the defenders. Washington, Seattle, Boston, Kansas City, and New Jersey are grouped with forwards.

There are two caveats to this analysis. First, it is preliminary. There are only 78 games represented and only Portland has a full set of data. Thus, more data could change the cluster behavior and this is only possible through WoSo Stats. Second, the clusters themselves still speaks more about each team’s defensive tactics and a team’s ability to cope against this pressure. More analysis needs to deconstruct the attacking styles and compare it to defending styles. In addition, if you recall, a majority of the conclusions made indicate the importance of location on the field.

Next time: analysis on passing direction and field location! Side note: there are only 40/80 games with field location data so please help WoSo Stats!

Add your thoughts, observations, and questions below. Make sure to check out WoSo Stats on Twitter, how you can help, the stats database, definitions of these stats, and some visualizations courtesy of WoSo Stats. Special thanks to those who logged the matches (the real heroes)! The visualization presented here are my own and will made public once all matches are logged. Find me on Twitter: @kryptobanana!