Jason McKenna recently looked at what happens when teams have ‘nothing to play for‘. Here he delves into more general trends around end-of-season performance to see what the numbers tell us
From Gameweek 30 commentators, managers and analysts alike frame it as the “business” end of the season. It is the time when seasons are won or lost, where European qualification can be decided and also survival in the Premier League.
But with these final crunch matches, is there a change in the style of play of teams?
In this article, I will analyse the attacking, defensive and pressing data of Premier League sides to analyse whether this time of the season sees a uniform adjustment in play styles to ensure maximum output towards the end of the season.
The analysis will also focus on the reasons for this and the possible need for a total rethink of managing player loads through campaigns.
How I will Measure Performance
The Gameweek 30 mark, with nine matches to play inclusive, was chosen as it is usually after the last international break of the season. This is a moment for teams that they can plan towards. With players that stay behind, there is an opportunity to train and work on tactical tweaks for those final matches of the season. But it is also a point where they can target specific fitness levels and output.
Six seasons’ worth of data, starting with the 2017/18 season and continuing through the most recent campaign, 2022/23, have been used to identify long-term trends. Looking at changes in the xG, xG conceded, Goals, and Goals conceded per 90 data before and after Gameweek 30 will help us understand whether there is a usual change in play style and output. It can go some way towards seeing if there is a uniform approach to how the end of the season is conducted. A conservative playstyle may be deployed to get the most out of the last few games. However, the opposite may be true, and teams may be fatigued, thus making more mistakes and playing in a more open fashion.
I have also included PPDA data as an insight into whether teams have changed their intensity of play in these last few games. A low change in this may indicate that teams endeavour to keep to a certain play style.
What the Data Tells Us
Overall, across the six seasons, teams stay relatively close to their PPDA output with only around a 0.14 pressing difference per 90 when comparing the period of Gameweek 1-29 and 30-38. This indicates that in this last quarter of the season teams still try hard to maintain their output in pressing style to see out a campaign. This is interesting, because it tells us that an effort at stylistic continuity supersedes, or at least is strongly determinant of, load management as the season comes to an end; it also tells us that even if teams have little to play for, their style is so inherent at that point that it does not alter materially.
However, the bigger differences come in expected goal data and actual goals output. Teams score more but also conceded more in these matches, with 0.10 more goals per 90 and 0.11 more goals conceded per 90 in these late campaign games, thus leading to a 0.21 goal swing per 90. The xG is also impacted, to a lesser extent, but the difference in expected goals and actual output shows that defences are less able to prevent goals in these final games as the 0.07xG change per 90 and the 0.09xGA data shows that attacks over-perform and defences under-perform through this period.
But the changes in output are also more felt wherever you are placed in the league table. Breaking it down into what I have classed as the main groupings there are big discrepancies in how these teams tackle the run into the end of the season.
The Top Two teams in the league tend to score less and concede more in these final matches of the season. This is also replicated in the expected data, which also sees a big swing. Overall, there is a 0.33xG difference per 90 for these top teams and a 0.38 actual goal difference per 90.
Teams fighting for European spots are also greatly affected, but their output seems to be a focus on the attack with an upward trend in xG per 90 and Goals per 90, but this high-risk methodology seems to be mitigated by the increased goals that are conceded.
| All Data Per 90 | xG Difference | xGA Difference | G Difference | Conc Difference |
| Top Two Sides | -0.17 | 0.16 | -0.10 | 0.28 |
| Sides 3-7 | 0.08 | 0.12 | 0.20 | 0.16 |
| Sides 8-15 | 0.07 | 0.05 | 0.03 | 0.02 |
| Sides 16-20 | 0.18 | 0.07 | 0.17 | 0.12 |
What is most surprising is that the sides 8-15th do not tend to have much difference in output in these runs into the end of the season. They are the group with the least variance from their pre-Gameweek 30 averages.
Another interesting find is that the sides in the bottom area of the table fighting against relegation seem to battle this with a focus on goals to stay up. They have the largest increase in xG of all groups, but again, this more open style of play is coupled with worse defensive data, and unfortunately, their poor-quality backlines concede more because of this, too. However, the marginal improvement of 0.05 goals per 90 may be the difference between staying up and going down for these teams.
The Possible Reasons For This
First and foremost, the most likely reason for the patterns reflected in the data above are due to fatigue and how teams plan around this. After huge seasons with large amounts of exposure through a campaign the last few games of the season are incredibly meaningful matchups that are being fought with tired bodies. This is why we see a downturn in expected and actual output as more mistakes are made. The literature around this states that “physical fatigue has a negative effect on the individual performance of soccer players”[1] according to a 2022 investigation. Fatigued backlines that are at the limits of their physicality allow opponents to score opportunities than they may otherwise have been unable to earlier in the season.
Another reality is that teams, especially those in the European fighting spots, will be trying to balance performances and injuries in the late stages of competitions. Players cannot play every game without a high risk of injury, so risks must be minimised. We, therefore, see high levels of rotation in these matches, around a 20% increase, and the introduction of five substitutes in recent seasons means that in game rotation is also a big factor. This also means that we will sometimes see lower-quality squad players getting an outing and those who are lacking a little bit of match sharpness, which again leads to anomalous performances and downturns in team outputs. It may also explain how teams can maintain their pressing levels, although perhaps suffer in the execution of that.
But the trend may also be accounted for in the periodisation of the season. The modern football club is so data-orientated that these issues may be accounted for and have been worked out in seasonal preparations. Periodisation is “the strategic planning of training to enhance performance for each competitive match across the season”[2], thus, clubs may have planned to peak at certain times of the season for their own individual goals. This is something that I opened out in a previous article for Analytics FC about Arsenal. Peaking at the “right time” is factored into periodisation, but also, being able to successfully see out a season is important. Each club has a different time for peaking. Relegation-battling sides may keep some efforts left over for the run-in to make sure that they can be fit for the all-important season-defining matches, whereas bigger clubs, who see the biggest defensive drop-off, may use a different approach. Their methodology to their planning may be to get points earlier in the season so that they can focus on multiple cup fronts come the end of the campaign.
This is again something that is present in load management and periodisation literature, which argues that it is a case by case basis of the individual plans of each club and their goals. There is no one correct way to plan things, but the differing levels of output between the areas of the league indicate these differing approaches and why some teams may see off-trend levels of output in this late stage. With this in mind periodisation “can offer coaches the opportunity to strategically manipulate training loads in anticipation of the difficulty of the upcoming fixture”[3] or fixture periods to target. A “weaker” side may look at a run of 4-5 games midseason that they believe they can get points from, whereas Cups are won at the end of the campaign, and thus “tapering”[4] performances are likely for this period, but the physical demands of fighting on multiple fronts means that the limited abilities of the players must be managed.
Football’s Long COVID?
Another factor may be the extreme post-COVID schedule players have been put through. Effectively, players have been allowed just one proper, extended preseason break since the COVID period, and the schedule has been dangerously hectic. In the data set that has been analysed, the two seasons before COVID appeared (2017/18 and 2018/19) saw much less pronounced differences in the data, with far more consistent outputs throughout the seasons. However, in the four seasons after this, the variation between the pre-Gameweek 30 output and the post-Gameweek 30 output is larger. This would tie into long-term, extreme fatigue, which organisations like FIFPro have been vocal about. There have been voices in the medical sphere of football that, even before COVID, have argued that schedules are too much. These will only increase in future seasons, but this period since the pandemic has seen the highest rates of football played in history. Contrary to the scientific warning of allowing players five days rest and recovery time between close appearances, the footballing authorities have increased the number of games. This led to the publication of the 2023 analysis by FIFPro where they stated that “the extreme levels of calendar congestion evident over the course of the season posed a pressing danger to the physical and mental health of players”[5]. It is, therefore, no wonder then that player output in the closing stages of seasons has become more anomalous than in previous years.
Conclusion
It is, therefore, clear to see the impact of end-of-season runs on teams. There are a multitude of reasons, but the evidence is clear that there are huge changes in the outputs of teams all across the table, and this is especially seen with poorer defensive outputs.
There are unforeseen changes where the top teams in the league have the “greatest” drop-offs, but these can be explained with some planning reasons behind them.
However, the overall message is that when a season comes to an end, players are at their physical limit. The analysis here gives an interesting insight into the effects on games, but the real message that needs to be taken away is how this overstretching of players is affecting their well-being for short-term injury issues and long-term health concerns. The dramatic drop-offs in this analysis for some teams should be a warning to authorities that this sort of enforced and unnatural levels of output cannot continue much longer.
[1] Dambroz F, Clemente FM, Teoldo I. The effect of physical fatigue on the performance of soccer players: A systematic review. PLoS One. 2022 Jul 14;17(7)
[2] Mark Read, Rick Rietveld, Danny Deigan, Matt Birnie, Lorcan Mason and Adam Centofanti “Chapter 10: Periodisation” in Calder, Alex. and Centofanti, Adam. (2022) PEAK PERFORMANCE FOR SOCCER. S.l: ROUTLEDGE
[3] Robertson, S.J. and Joyce, D.G. (2015) “Informing in-season tactical periodisation in team sport: development of a match difficulty index for Super Rugby,” Journal of sports sciences, 33(1), pp. 99–107
[4] Pyne, D.B., Mujika, I. and Reilly, T. (2009) “Peaking for optimal performance: Research limitations and future directions,” Journal of sports sciences, 27(3), pp. 195–202
[5] FIFPro, “Extreme Calendar Congestion: The Adverse Effects on Player Health and Wellbeing”, PWM Annual Workload Report- Men’s Football (2022/23Season)
Header image copyright IMAGO/ActionPlus/David Blunsden
