In his debut blog for OptaPro, data scientist @WillTGM analyses central defender passing trends in the Premier Leag…… 17 Jan

VACANCY - we are recruiting a data scientist, who will work across @OptaPro & @OptaSuit. Opportunity.… 16 Jan

Fantastic to have @WillTGM join the team. We'll be publishing Will's first blog for OptaPro this week, which explor…… 14 Jan

Valuing passes – @GarryGelade considers how passing performance can be evaluated using a Pass Reward model. Vision.… 10 Jan

The OptaPro Blog

RSS Feed

Welcome to the OptaPro blog, featuring news and analysis from OptaPro's cutting-edge research team.

BLOG: How PSV changed their playing style and got back to winning ways

PSV Eindhoven’s 3-0 victory over nearest rivals Ajax last month secured them the 2017/18 Eredivisie League title with three matches still to play. Their success is an emphatic response to a disappointing 2016/17, when they could only secure third place in the league, six points adrift of winners Feyenoord.

In this blog, adapted from a presentation delivered at the 2018 KPMG Netherlands Sports Analytics and Sports Technology Conference, we apply Opta data to analyse how PSV changed their style of play to regain the Eredivisie crown.

A summary of the 2016/17 season

PSV finished 2016/17 with eight fewer points and 20 fewer goals compared to the previous campaign. However, their defensive performance actually improved. PSV conceded just 23 goals, nine fewer than the year before. In fact, they only conceded 80 shots on target, the lowest number recorded in the Eredivisie since Opta began data collection in 2010/11.

Defensive Summary: Top-5 Eredivisie 2016/17

Eredivisie Defensive Summaryv2

Despite the lack of goals, PSV were actually still creating high quality opportunities, but failed to convert these chances. In contrast, champions Feyenoord scored 17 more than their projected xG total.

Attacking Summary: Top-5 Eredivisie 2016/17

Eredivisie Attacking Summary

What perhaps embodies PSV’s struggles was the drop in goal output from Luuk de Jong. After scoring a career best 26 league goals in 2015/16, he could only manage eight goals in 32 appearances, significantly underperforming against his xG total. Including penalty attempts, he recorded the poorest goal to xG ratio captured by Opta during the last five years for any player across Eredivisie and the ‘big-5’ European leagues.

Luuk De Jongv2

PSV’s 2016/17 playing style

PSV’s playing style from 2016/17 featured the following characteristics:

Possession-based approach: After Ajax, they had the most possession in the league at 59%, an increase on the previous campaign.
Emphasis on crosses: On average they were delivering over 18 open play crosses per game, PSV’s highest under Phillip Cocu.
High winning possession line: Their average line for winning possession was roughly on par with Feyenoord, with only Ajax commanding a higher average line.
Average attack speed: Regardless of the game state, PSV appeared to lack urgency in their direct speed of attack. Across the Eredivisie, they sat in mid-table for attack speed, no matter whether they were winning, losing or drawing.

Their ever-consistent speed of attack, irrespective of game state is interesting, because it suggests one of two things: either PSV had complete confidence that their style of play would result in victories, even if they went behind, or that they didn’t have a plan B.

In most cases, the direct speed of attack goes up for both winning and losing sides.

Key players involved in sequences leading to goal attempts

Using the Opta sequence framework, we can establish which players are regularly involved in a possession sequence which results in an attempt on goal.

For example, Marco van Ginkel was involved in eight shots (per 90) when PSV were drawing, 5.4 when they were winning and 5.7 when they were losing.

In contrast, as shown in the table below, when PSV went behind Gaston Pereiro and Jurgen Locadia became more involved. When they went ahead, the player most involved was Steven Bergwijn.

A more detailed summary of the sequence framework is available here, in an earlier blog by Sam Gregory.

Players involved in 2016/17 open play shot sequences: arranged by game state

Table 1

Average per 90 minutes. Only players with a minimum of 750 minutes played included

Key Changes for 2017/18

The six players highlighted above in blue left PSV during or after 2016/17. Four of them, Jetro Willems, Héctor Moreno, Davy Pröpper, and Andrés Guardado, were integral members of the squad, with each having played over 2000 leagues minutes during the campaign.

During the summer of 2017 PSV recruited three new first-team players: Derrick Luckassen from AZ Alkmaar, Hirving Lozano from Pachuca in Mexico and a new back-up goalkeeper, Eloy Room.

Luckassen has been a fringe player this year, starting only nine games. Given that four of the players who left the club were key players defensively, it may not come as a surprise that despite winning the league, PSV’s overall defensive performance has dropped.

PSV conceded 39 goals, the most conceded by a Dutch league champion since 1984/85. Their xG total was even higher, at 41.5, which is their highest season total since Philip Cocu took charge in 2013.

PSV Eindhoven Defensive Summary: Last 5 Seasons

PSV 5 Year Defensive Summary

When we compare their attacking output to the previous season, we can see that PSV enjoyed a small increase in the number of attempts at goal and xG, however the main reason for their success was a massive swing in terms of their conversion of chances into goals. They went from underperforming on xG by nine to overperforming by eight – a positive 17 goal swing.

PSV Eindhoven Attacking Summary: Last 5 Seasons

PSV 5 Year Attacking Summary

Changes to their playing style

Having had a possession-based approach in 2016/17, there was a noticeable drop (9 percent points) in the amount of possession PSV had last season.

In addition, the volume of open play crosses they delivered into the box dropped by 25% and in games when the scores were level or when PSV went behind, they had the highest speed of attack in the league.

One of the reasons why crosses went down was because Jürgen Locadia was being utilised as the main striker during the first half of the season, but the change in the speed of attack, coupled with the fact their possession-winning line was significantly lower, suggests that PSV have made a conscious decision to give up control of a game in order to make a game more open – and allow their attacking players to counter-attack and hurt the opposition.

One thing that stands out is if PSV took the lead or went behind in a game, the midfielders had less involvement in the sequences. In comparison to last season, when five of the top seven were midfielders, only two appeared in the top seven this year when PSV were level. More interestingly, it seems that when they went behind, one of their key players involved in chance creation was their right back, Santiago Arias.

Players involved in 2017/18 open play shot sequences: arranged by game state

Table 2

Average per 90 minutes. Only players with a minimum of 750 minutes played included

Using Arias as an attacking force

During 2017/18, Arias created more chances than any PSV player from open play, completed more passes in the opposition half and completed the most passes which entered the final third.

As highlighted by this graph, it is clear that irrespective of game state, Arias was involved in more instances leading to a shot compared to his team average. Arias is highlighted in blue, with the red line being the team average.

Arias Screenshot

It is also worth noting that in addition to being a vital attacking asset, Arias also had solid numbers defensively too. He was second to Joshua Brenet in total recoveries and was also in PSV’s top three for tackles and interceptions.

Will PSV retain this style of play next season?

The 2017/18 incarnation of PSV relied more on a counter-attacking style, creating a less compact playing area, with Joshua Brenet and Santiago Arias being key in both defence and attack.

Whether they retain this approach next season will depend on whether their key personnel stay at the club this summer.

If PSV do find themselves with the need to replace key players, players who have played in teams with similar styles to the two most recent versions of PSV may be natural fits. These teams can be identified using a statistical approach that finds groups and similarities between a range of teams based on their relative performance in some key stylistic attacking metrics.

Using this type of analysis, some European teams who adopted a similar attacking approach to PSV this year (quick in transition, quite central) include Young Boys, PEC Zwolle, Zulte Waregem and Brondby IF. On the other hand, sc Heerenveen, Club Brugge and FC Nordsjaelland are amongst the clubs from 2017/18 whose style most resembled the PSV attacking approach of 2016/17 (more possession-based, utilising the width of the pitch and looking for crossing options).

Either way, it will be interesting to see which approach they take and from their rival’s perspective, will they look to change their own approaches to try and wrestle the title from PSV’s grasp?  

Things will become clearer no doubt when we see which players stay and which new players come in.


Posted by Michiel Jongsma at 00:00


Post a comment