This piece first appeared on Ben’s site ‘Spreadsheets for Goalposts’ here
Load management is a key balancing act for any professional or amateur athlete. It’s therefore not a surprise that professional sports teams employ highly educated and qualified specialists in their sports science and performance management teams. Unfortunately, I am not one of those, so instead here is a case for increased rotation of key Reading FC players from an (enthusiastic) amateur…
Covid means the 2020/21 season is unique – there are 238 days between the delayed season opener and regular-season final fixture, compared with 275 days for the last full normal season (2018/19). That shakes out as a game every 5.2 days on average, meaning almost a day less recovery per game compared with the ‘normal’ game per 6 days.
This is further compounded by the reduced summer break of 52 days, compared with the 90-day break between 2018/19 and 2019/20. On the face of it, this is a headache for managers, and you might reasonably expect a high amount of rotation. Bearing in mind the injury record, especially to key players (significant absences for Moore, Yiadom, Swift, Méïté, Pușcaș and shorter absences for Richards and João) I expected to see that we have had to use a large squad to manage these absences.
But that hasn’t happened. Compared to recent seasons, Reading have used fewer players overall – the 24 that have featured in the Championship this season is the lowest of any side in the league, and a significant drop from the 31 that featured last season. Given the circumstances, I had expected ‘rotation’ to be high across the league, but interestingly the number of players used so far by the median team is currently one or two lower than for previous (complete) seasons.
I can think of two possible reasons: (i) an unfair comparison as the numbers for previous/complete seasons have been boosted by giving opportunities to fringe players in end-of-season dead rubbers, and (ii) reduced transfer activity as a result of Covid finances – 2020/21 transfer spend was around 1/3 of the level of the previous two seasons.
It’s not only the number of players used, but also the concentration of minutes too. 11 players have played more than half of the maximum possible number of minutes – a level comparable with previous seasons, but around twice as many players as usual have played a very high share of minutes, with seven playing 75% or more, and four playing 90% or more of minutes.
So why does this matter?
I’ve already touched on the truncated schedule, but should this impact professional, elite athletes? Overall, while football players are probably fitter now than ever, so the physical demands are greater – so players likely face similar levels of fatigue each game as they previously have. But the compacted fixture list is the new element and might reasonably be assumed to cause higher injury risk. Digging into the literature, several papers have explored the factors behind injuries in football. Here are some brief summaries of papers that seem applicable to the current scenario:
- Accounting for games played, Bundesliga players got injured at over three times the usual rate after the 2019/20 Covid break (Seshadri et al, 2021)
- 72 hours is insufficient to fully recover following a match (Silva et al, 2018)
- Injury risk during matches is around 10x the injury risk from training, and the risk of injuries increase at the end of each half (Lopez-Valenciano et al, 2020)
- Injury risk increases during a congested fixture period (Dellal et al, 2015)
- Player positions and roles have varying degrees of intensity – strikers and wingers on average sprint twice as far as centre midfielders each game (Andrzejewski, 2013), while full backs and wide midfielders perform significantly more turns (Baptitsa, 2018)
So why does this matter now?
Well we happen to be coming into the most congested part of an already congested season. Plotting the number of fixtures in a rolling 12-day window we can see we have just started the second of two blocks where this 12-day window contains between three to four games. At four games spread evenly over 12 days, we can’t expect players playing a high share of minutes to be performing at their optimal fitness given the need for >72 hours to fully recover.
The below chart plots this compared to last season, and 2018/19 as a ‘normal’ season. We can see a 12-day window containing four games is usually rare and limited to the Christmas schedule in a regular season. The resumption of the 2019/20 season saw a congested fixture list of three to four games every 12 days, but 2020/21 has three clear periods where 12-day periods regularly contain three to four games, and we are just starting the longest one. For reference, 2018/19 contained two 12-day periods containing four games, last season contained eight (seven after restart), but this season is set to have 25 occurrences, or 12 times the ‘normal’ amount.
To help visualise the build-up of fatigue compared with previous seasons, we can make an overly simplistic representation for athletic ‘load’. If we say each game represents +100 ‘fatigue points’, and each rest day reduces this by 25 points (meaning four rest days is sufficient to fully recover from a match) then we can see how the condensed fixture might build up fatigue. By aggregating the load from these fixtures and the impact of rest days, again in 12-day windows, we can plot how this build over the season:
The maximum 12-day ‘load’ score here in 2018/19 was 225 ‘points’. In 2019/20 it was 350, but this season it peaks at 425. There are 82 days this season where the ‘load’ is at or above the 2018/19 peak of 225 – equivalent to one third of the season. Right now (the dashed line) we are approaching the second period when this metric peaks, but the unique element this time round is after this peak; it remains elevated between 200-300 points, whereas after the December peak (days ~90-120) it returned to 100 or lower.
Put another way, we are approaching the period where the stresses and strains on players are expected to be over double the levels usually seen in a season, and it will persist for the rest of the season.
But is this meaningful?
If we look at the occurrences of injuries to key players, we can see we lost four key players as my simple metric of ‘load’ increased around Christmas, with John Swift the only key injury occurring at a lower ‘load’. While I don’t believe this approach will have any meaningful power to predict injuries, it is striking that we are now approaching a similar level of ‘load’ and could well be facing more injuries in this spell without rotation.
Indeed, we should expect injuries to occur when ‘load’ is higher simply because more matches means more opportunities for injuries to occur. But nevertheless, I think it neatly displays why the need for rotation is a concern right now and raises a question as to whether some of those injuries could have been avoided during the December peak.
Of course, the sports science and performance analysts will have a significantly better handle on this than someone like me, and I imagine (and hope!) training will have pivoted to primarily recovery and tactical focuses, with a lower emphasis on intensive fitness training. Likewise, individual players will have different abilities to manage workload, and again I would expect the relevant experts to work within these different limits.
But it remains striking that Veljko Paunović has been resistant to rotate players unless it is forced by injury. My case for this is via the high share of minutes for few players (as shown above), the late use of substitutions (at 17 mins, our average substitute run-out is the second lowest in the league) and leaving unused substitutes (156, the fourth highest in the league). My concern is that recommendations for resting certain players may be going ignored, with the manager making the final say on team selection and prioritising the immediate term success of the team.
As an aside, I think it’s likely players are aligned with this approach. It is very rare that a player would opt to play anything less than every possible minute, even if it damages their long-term fitness and career prospects. There are several cases of players never recovering from continuous bouts of injury because they are too important to rest (Owen, Torres, Bale and all eyes on Kane for the next big one).
Deeper dive by position
We can explore the lack of rotation by looking at how many minutes each player has played in the latest 20-game block (since the November international break), broken down loosely by position, so the charts aren’t so busy. I’ve excluded players who have played less than 100 mins in this period.
Liam Moore’s injury came at the end of the first period of extreme ‘load’, when he had played every minute – since then Tom McIntyre has played every minute, while Michael Morrison has only missed the cup game against Luton Town, when the squad was rotated.
Omar Richards, another player to suffer an injury during the first period of ‘peak’ load, has returned to fitness, and I was glad to see him rested against Millwall, even though I suspect this was a tactical omission due to the opposition’s aerial threat. Tom Holmes has played a large share of minutes, but also benefitted from the occasional rest (he has only played >300 mins twice out of a possible seven occurrences).
Tomás Esteves has barely featured in recent weeks after being more heavily involved over the Christmas period, and Andy Yiadom has only just returned from a long layoff. It appears there has been some (deliberate or enforced) rotation at right back, possibly reflecting strong competition from three options, but I would hope Richards is protected as a key player and rotated in coming games.
Josh Laurent and Andy Rinomhota have been virtual ever-presents throughout this period, each only missing a handful of minutes (again the Luton cup game on January 9), with the former missing the win over Bristol City thanks to an ‘accumulation of knocks‘. John Swift came back from injury, played a lot of minutes in late January, and subsequently got injured again. Alfa Semedo has been a regular in different positions and has looked a lot more comfortable as part of a double pivot, which hopefully provides an option for rotation with both Rinomhota and Laurent in the coming weeks.
Ovie Ejaria has played nearly every minute over this period, while Michael Olise has generally been managed with more regular rest until the most recent run (three full 90s and two 87’ subs in the last five games). Sone Aluko was being used as a regular option over the Christmas period, but has been sparingly used in recent weeks.
This is the position of largest concern for me, as I believe having at least two of Ejaria, Olise, Swift and Méïté is key to our attacking threat and losing Olise or Ejaria (on top of the other two) could seriously damage our prospects. I would be looking to rotate them out in the coming weeks to keep them as fresh as possible for the run in.
Lucas João is another player who is rarely substituted or rested, despite his reputation as an injury-prone player. He missed a chunk of the season after getting injured in December (again at a period a peak load when he wasn’t rested), and since coming back to fitness has been ever present in recent weeks. Yakou Méïté featured heavily over December, before also suffering from injury, while Sam Baldock has only played significant minutes in the absence of João.
With João being such a key player to our style, it’s imperative we give him every chance to be fit for the run-in, and I would be keen to see him subbed much earlier, or starting on the bench as an impact option in the coming weeks. George Pușcaș returning to fitness will hopefully provide an alternative that provides João some rest. Unfortunately, Baldock, Aluko and the youngsters haven’t had many moments of brilliance this season, but I believe at some stage they become more of a threat than our exhausted and overused attacking core.
Pauno’s balancing act
I accept Paunović is stuck between a rock and a hard place. If he rests or substitutes a key player and we don’t get a good result, the blame falls directly on him, even if the longer-term prospects are improved by having a fitter squad for future games. Indeed, some criticism was directed at him after substituting Holmes off for Yiadom against Millwall shortly before they turned the game around, largely unfairly in my opinion.
When he does rest players, there is no counterfactual to compare against – we might have picked up more points by playing the best 11 players every game thanks to extremely good luck and no injuries… but unfortunately we can’t prove this without a time machine. Finally, even with huge leaps forward in the data available to the sport science team (through innovations in wearable sensors and/or camera tracking), injuries remain extremely difficult to predict.
This season is unlike any previous seasons in terms of the lack of recovery between games. It’s likely that players are more fatigued than usual and managing the injury risk for key players will be a challenge for Veljko Paunovic. So far, key players have played pretty much every minute they are available for, and I’m concerned we may see another batch of injuries in the coming games unless a different approach is adopted.
Unless the sports science and performance analyst teams have strong evidence that these key players are able to manage this period, I would be seeking to rotate key players in the coming weeks, especially the likes of Lucas João, Ovie Ejaria and Michael Olise due to the intense physical requirements of their positions and large share of minutes played in recent weeks. I fear without this rotation we may limp into a play-off campaign missing key threats.