Reading’s Defence and Valuing Defensive Midfielders

Reading’s Defensive Numbers

After first going through the WSL data last summer, Reading’s defensive numbers were something stood out. Unfortunately, I can’t remember the exact numbers that stood out, but it was about them being hard to play against, particularly with their pressing and narrow midfield they often used last season.

I haven’t watched them closely enough to talk at length about them, but they still have some interesting numbers in 2019/20.

Playing out from the back

To start, they don’t make many pressures. They make the 3rd lowest number in the division, but do have the 5th highest percentage in the final third, suggesting a more front-footed approach. Adding support to the front-footed approach, Reading’s opponents have the smallest average possession duration so far this season.

Then, for possessions that begin in the opposition’s defensive third, they’ve conceded the smallest percentage to have more than three passes by quite a significant distance. 26.0% of possessions that start in Reading’s attacking third don’t manage three or more passes, the next lowest is Arsenal with 33.2%. There’s roughly the same gap between Reading in 1st and Arsenal in 2nd as there is with Arsenal and the 10th placed team.

Watching some clips, I thought I’d found the reason. For some possessions starting with the ‘keeper, a new possession might start as it progressed from being from the ‘keeper to regular play.

So I decided to change it to remove possessions that start with the ‘keeper to see what happened. While there was some difference in the rest of the league, Reading remained almost the same. Now 27.6% of possessions didn’t manage more than three passes, still some way ahead of Chelsea, who moved up into 2nd place, with 32.7% of possessions.

However, what if Reading’s opponents don’t spend a time a lot of time on the ball or make many passes in possessions starting at the back because they’re able to progress the ball forward quickly? You’d associate those two things with a team difficult to play against, but it’s not impossible.

Looking at how often team’s progress the ball from front to back, there’s nothing particularly conclusive. They’re 5th for percentage of possessions that start in an opponent’s defensive third and reach the final third, which is the same as their league and xG Difference standing, so it doesn’t jump out as unusual. Sure, after the other numbers it might be slightly disappointing, but it doesn’t raise alarm bells.

But then things did start to slowly get a bit more interesting. Not including passes from the ‘keeper, Reading’s opponents had the highest percentage of passes that attempted to move the ball out of their defensive third. Again, it’s worth questioning this though. What if this is because they go long due to being under pressure?

Turns out, looking just at passes that successfully move the ball out of the defensive third, Reading’s opponents had the lowest success rate in the league.

All of this seems to paint a picture of a team who makes it hard to progress the ball against. But to throw another spanner in the works, Reading concede the 6th most shots from possessions that start in the opposition’s defensive third and have the 2nd highest average xG for these possessions (the possession could contain more than one shot which could skew it slightly).

It seems like the old pressing team cliché is true for this Reading side. They might be difficult to progress the ball against, but if you do manage to get past them, you could get a decent quality chance out of it.

Are Reading vulnerable in transition?

This is only looking at possessions that start within the opposition’s defensive third though. I wanted to try and find out how teams fare when in transition. For the most part, possessions that start in the defensive third feel like they’ll fit more into a build-up phase. Reading have ample time to get in shape and look to cut off passing lanes. And this where hacking things together with lots of assumptions begins.

I’ll be transparent as possible about how I worked these kinds of things out because I fear I might have missed something or used some silly logic. To try and find out how vulnerable teams are in transition, I wanted to see shots they conceded within x seconds of a turnover (x was a hard number to decide, too short and there’s hardly any shots, too long and it’s basically just all of a team’s xG). So I made a dataframe that only had successful turnovers (tackles/duels, interceptions, ball recoveries and passes that were of type interception/recovery) and shots.

From there I selected shots where the previous event in the dataframe was a turnover by the team who took the shot and was no more than 10 seconds before the shot. I also decided to say the turnover had to have happened in the first two-thirds of the pitch. After not putting this in, there were quite a few shots where a player would make a ball recovery following a header clearing a cross, which isn’t really what I was after.

Looking at the xG per game conceded from these shots, Reading have the worst numbers in the division. It’s fascinating how they can post such good numbers for making it hard to progress the ball, only to have such poor numbers here.

They give up the 7th most shots from transition attempts, not too worrisome, but have the highest xG per shot by a distance. The average shot they concede following a turnover has an xG of 0.19, the next closest is Birmingham with 0.14. There are only three teams in the division to have an xG per shot in double digits for these shots, showing just how much of an outlier Reading are.

I thought I’d go a step further too. Following the same logic as above but with progressive passes rather than shots. This time I looked for progressive passes (those that move the possession at least 10 yards closer to goal, which is easier done in this instance because it’ll usually be the beginning of the possession) that came following a turnover.

I went through the same process but this time rather than 10 seconds, the progressive pass had to come within the time of the turnover + the duration of the turnover + 1 second for rounding, to try and ensure it was directly following the turnover – as well as being the player who regained possession. So a player wins the ball back and then sends their team on their way.

For this, Reading didn’t fare too bad. They conceded the most progressive passes per game following a turnover, but when looking only at completed progressive passes, they had the 6th best numbers, not great but not concerning either.

Extending it further, changing the duration + 1 second to 10 seconds and getting rid of the rule that it has to be the same player, Reading fare better and have the 3rd best numbers for both the number of completed passes against and pass completion of these passes.

Again, they’re not constantly being torn open on the transition, but when it happens they’re giving up enough high-quality chances to concede the most xG from them.

Example vs Bristol City

From what I have seen of Reading, they seem to play direct and then press to try and win the second balls. If it comes off they can hit the opposition in transition, but if it doesn’t they can get hit in transition themselves. Players can quite easily get caught ahead of the ball and leave the opposition with space to run into ahead of the defence.

Their second goal conceded against Bristol City shows some of this. It starts with a throw-in, which has a big impact, but you can see 6 Reading players in shot in a fairly small area of the pitch (maybe 8 depending on how big of an area you class as small).

Reading gain possession, but once it’s back for Bristol City, there’s plenty of space on the opposite flank. They play around the press more than through and progress into the opposition half.

Bristol City lose the ball again and you can see 5 Reading players in shot all push forward, but it means once City regain possession again these 5 Reading players are caught in front of the ball with 3 rushing back to try and defend and (I think) only Jade Moore there to try and protect the defence.

In the end, it’s a great pass from Charlie Wellings to play in Ebony Salmon in behind, but as soon as the ball is played to Georgia Wilson the defence is exposed and has a lot of space behind them. Wilson could have even gone inside to Carla Humphrey, rather than outside to Wellings, who could have turned to see both Ebony Salmon and Yana Daniels making runs in behind in what looks like a 2v2.

The narrowness seems to be a Reading thing too. They have the lowest average width (looking at the difference in the maximum and minimum point of possession horizontally) for possessions with more than three passes, so if the ball is turned over there does seem an opportunity to switch the play.

Conclusion

To be fair to Reading, they’re having a decent season. They’re 5th both in the table and xG Difference. However, if they can improve in transition it could be even better for them. For reference, Reading concede 0.36 xG per game from the transition definition I hacked together, Man United, one place above them, concede just 0.09 per game.

If Reading think they gain enough in attack to make up for their poor transition record, viewing how they play as a high risk and high reward strategy, it’s questionable. Reading do create the 4th most xG per game from transition chances, but only create 0.03 more than Man United. I’m only using United as a comparison because they’re 4th and 5th and the only sides outside the top three to have a positive xGD, but Reading’s approach sees them gain little in attack but lose a lot in defence compared to their rival for 4th.

It may be that Reading are going through a bit of a transition in general. They lost the trio of Kirsty Pearce, Rachel Furness and Gemma Davison over the summer, all of whom were experienced and important in the spine of the team. They’ve given minutes to some younger new signings like the Norweigian pair Kristine Leine and Amalie Eikeland, as well as other new signings or players who may not have played as much last season, like Angharad James and Sophie Howard.

As I said, I need to watch a lot more of Reading before making saying anything definitive, but their numbers continue to be interesting and how they play can be quite fun to watch from a neutral point of view.