Mind the gap

Guest blog by Andy Watson @andywatsonsport

Which is the hardest step-up to handle in the English football leagues?

When attempting to assess whether a player is going to be able to perform well for your team it is important to understand the current level of competition that the player is performing in, and how that compares to your own league. It is common sense that a player playing in the third tier of Belgian football will not necessarily be able to reproduce the same levels of dominance in the Spanish Primera Division.

The intention behind researching performance at different levels of the game then is to attempt to reduce risk and give greater confidence to a scout’s assertions on a player, or the predictive data that is already within the recruitment system. There are an almost infinite number of factors within whether a player might be able to perform well at a higher level that it can never be an exact science. However, the more research that is undertaken should ultimately reduce the risks involved.

To begin with the research is centred around understanding the step-up in the English leagues. There is available data down through the league system within England going back a few years now and that was a key consideration. Also, before diving into individual player data it made sense to try to understand how teams fared across the different levels of the league system to calculate some baseline “quality gaps” between the divisions. These can then be used in future assessments of individual players to understand their own performances within a specific context.

Obviously, football works on a hierarchical system, and England is no different. The team who gains the most points in a season will be promoted to the next tier in the league system. They will often be accompanied by at least one other team who is promoted by other methods across the English football pyramid.

It then stands to reason that a side that achieves promotion will then be competing alongside teams that are of a higher level the following year. Therefore, if you compare performance data across the two seasons you would expect a certain level of drop-off from the promotion season to the new season. Indeed, many pundits/observers say before the new season begins that promoted sides will be happy to finish one place outside the relegation zone, intimating that survival should be the goal for these teams.

Analysing Promotion and the Following Season

Nevertheless, it is commonly accepted that some leagues are easier to be promoted into than others. I ran a poll on Twitter to ascertain the general opinion of my followers as to which step-up is the most difficult to handle within the professional English game.

The vast majority of respondents identified the promotion from Championship to Premier League as the most difficult to handle. This was certainly the result I expected, and it is just as interesting to see that the step-up from League Two to League One got zero votes in the informal poll.

Are the respondents to this poll correct? Moreover, is it possible to calculate the “quality gap” between the divisions as this would be important in future in assessing the prospects of a player playing in a lower league being able to adapt to a league higher in the league structure.

The idea of using promoted teams to measure this in the first instance is because it reduced certain variables. For example, most teams who are promoted retain the same manager/coaching staff for at least some of the following season, so we can expect a level of consistency in approach that a change in regime doesn’t usually grant. Also, there are no parachute payments to consider when moving up the divisions, though obviously there will still be an increase in budget as clubs anticipate the greater revenues available in a higher league through gate receipts, TV revenue and sponsorship opportunities.

The most basic measure of the difficulty of the step-ups across divisions would probably be to look at the ratio of teams that are relegated directly after promotion, or even the average position of promoted teams the following season. When you run the numbers along these lines these are the results over the last nine seasons:

This rudimentary use of historical information back up the result of the poll and also suggests that the gap between being promoted to League One compared to being promoted to the Championship and the Premier League is a fairly substantial one. The mean position of promoted clubs into League One is almost exactly mid-table and no sides were relegated at all over the last nine seasons. This looks set to change this season as Tranmere are currently looking likely to suffer that fate.

This is certainly enough information for the lay person and is a fairly good guide to results that you’re expected to get. The other piece of good news about doing it this way is that there are decades of data across the EFL to get stuck into. The question of how relevant the results become after a few years though is a valid one.

Would Analysing Performance Data Add Clearer Insight Into Post-Promotion Difficulties?

However, those historical results don’t necessarily give the measurement of quality gap that was wanted at the outset. With the advances and widespread nature of football analytics in the modern-day performance data is available from promotion seasons and can be used to compare to the performance data in the following season in the higher division. This means there are specific areas of a team’s performance that can be compared. 

The data so far only goes back four or five seasons so the sample sizes aren’t huge. Nevertheless, the analysis throws up a number of interesting areas for debate.

Attacking Play

Unsurprisingly, teams that are promoted find the next year more difficult at both ends of the pitch. The mean has been taken across all instances of promotion could be accessed from Wyscout. 

It is quite clear to see the differences between the levels. Being promoted to the Premier League means that a club scores almost 40% fewer goals per game and concedes over 60% more goals per game than they did in the Championship. Interestingly though the xG figures tell a slightly different story, at least in the attacking data. The promoted teams that have played in the Premier League and the Championship actually have very similar drop-offs in xG results. The apparent difference is converting these similarly rated chances to goals in the Premier League. The shot volume drop-off is also very similar across the Premier League and the Championship.

The biggest problem that teams encounter after promotion to the top two divisions is conceding so many more goals and so many good chances. It is an obvious thing to say but the teams that are in a higher division have greater quality going forward, and again this is accentuated in the top two divisions and in the Premier League in particular.

Of course, there are always teams that perform better than others. It will come as no surprise that the two teams that have out-performed almost all other teams promoted to the Premier League are this season’s Sheffield United side and last year’s Wolverhampton Wanderers. Sheffield United are the only side to have been promoted and conceded fewer goals per 90 minutes than in the Championship. Most of the other Sheffield United metrics are also excellent in comparison to the others and sit alongside Wolves’ from the 2018/19 season.

Meanwhile, there is one stand out team when observing the League One to Championship promotion, Millwall 2017/18. Though they did score fewer goals in the division above they also conceded a massive 35% fewer in their Championship season than in their League One promotion campaign. Looking deeper though there was a massive difference between their goals conceded and their xGA, a 61% difference to be exact, which obviously is huge and hints at an overperformance in that season.

Passing & Possession

Looking more into the build-up and performance differences in promoted clubs it is clear to see the problems that promoted teams face in a higher league. Again, the differences in the gaps between leagues is clear as well.

Interestingly, League Two clubs enjoy a greater number of passes and a higher level of accuracy in the division above. This is decidedly not the case in the Championship or Premier League.

When a team loses 10-15% of their number of passes and also possession it must be a real culture shock, at least in the first instance, to that team and to the management. Even teams that stick to their philosophies and their strengths will find their capacity to play being diminished. Examples of this that spring to mind are this season’s Norwich City and Barnsley teams in the Premier League and Championship respectively.

Norwich have seen their number of passes drop by 11% and their possession percentage drop 13% despite ostensibly sticking quite closely to their ideals. Obviously, their results leave them currently at the foot of the Premier League so in essence that hasn’t worked, despite some positive performances and general media consensus of their team not being a bad one.

Barnsley are in a similar predicament in League One, and their data tells us that their passes are down 15% and possession is down 11%.  These figures for Barnsley are actually greater than the mean from the study so maybe there has been either a slight change of emphasis, perhaps under the new coach, or maybe the team has just suffered against the quality of opposition more than average.

The teams that don’t have as extreme drop-offs in passing performance are the ones who had a more counter-attacking or direct style to achieve promotion. The fact that there are relatively few of examples of these in the past few years speaks to the general effectiveness of the strategy and the changing nature of football in this country, but when it is done well it can still work. The obvious examples in the Premier League are Brighton 17/18 and Cardiff 18/19. They saw very little change to their passing metrics. In the final results, survival was gained by Chris Hughton at Brighton and it could be argued that Neil Warnock’s Cardiff side slightly outperformed expectations in the Premier League despite immediate relegation.

From League One to the Championship two sides stand out as not losing much in their passing metrics, Blackburn 18/19 and Sheffield United 17/18. This is possibly more about proven ability of the players within those sides at a higher level and the ability of the coaching team to use their abilities on the pitch than those sides being direct. In fact, Sheffield United 17/18 had by far the highest number of passes per 90 (427) than any other promoted side in the study.

Progressive and Smart Passing

Passing and possession do feed into the type of passing that teams are able to accomplish but it is perhaps worth highlighting the difficulties that stepping up in level causes for progressive passes and smart passes.

Smart passes are defined by Wyscout as:

Something more than a simple pass, not so easy to be done. There has to be some idea in the pass, something creative, when the player is cutting the lines and winning some advantage for his teammates with this pass, leading them in good position to attack. 

Clear? Well, regardless of the minutiae of the definition, and the difficulty in identifying them(!), it is clear that all promoted sides produce fewer of these passes in a higher league. Yes, they have less of the ball in general, as demonstrated above, but there is a much lower percentage of smart passes attempted than possession lost in promoted sides.

Again, the difference is more pronounced in the Premier League and Championship but, interestingly, this is a particularly difficult statistic for the League Two to League One teams to keep up after promotion as well.

Pressing

A little bit more jargon to deal with in this section as we look at the differences in passes per defensive action. PPDA is calculated by Wyscout by dividing the number of opponents’ passes allowed by the number of defensive actions taken (duels, tackles, interceptions, fouls). This is calculated for actions taken within the final 60% of the field, so the opponents’ defensive half, plus a little bit more.

Essentially the higher the number in PPDA the more passes have been allowed without a challenge. Manchester City, for example, usually have a low PPDA figure because they will often attempt to win the ball back early. Newcastle, at the moment, will have a high PPDA because of their tendency to sit in a low block and wait for the opponent to come to them.

What the numbers in the graphic demonstrate, however, is that promoted teams don’t win the ball back as often, or even attempt to win the ball back as often or as quickly as they did in the previous season. In the Premier League this is multiplied by a factor of two over Championship promotions.

The promoted sides are also pressed a little more in the higher divisions, though there is little difference in this between League One to Championship compared to Championship to Premier League.

Again, it is Wolves 18/19 and Sheffield United 19/20 that perform the best in PPDA metrics. Both clubs still allow more passes per defensive action against them, but keep it down below 3 extra passes. United also buck the trend of being pressed more in the Premier League, they actually have a higher opponent PPDA in the Premier League than in the Championship.

Conclusions

This study is being undertaken with a view to understanding and, ultimately, measuring the performance gaps between leagues. Without a doubt the performance of the promoted clubs over the last four to five years has demonstrated that the gap between the Championship and the Premier League is the widest. This is similar across most metrics, though the gaps change substantially from some metrics to others.

The quality gap between League Two and League One is quite small and it is easy to see why the vast majority of promoted sides from League Two go on to thrive, let alone survive, in League One.

The data suggests that the greatest gap between the Championship and the Premier League lies in the number of goals that promoted sides concede in comparison to their time in the league below. This perhaps rather simply demonstrates the nature of elite performance sport in that the gaps at the top are greater than lower down, as in a normal distribution of almost anything in society.

However, there is room for further study in this topic, especially for managers, coaching staff and recruitment agents. The amount of possession lost, the reduction in passes and the changes to PPDA are ones that most people would recognise from the stands and they make complete common sense. But to be able to distil it down into how big the holes are and then think of potential ways to close that gap either on the training pitch or in the transfer market.

The next step on the research would be to carry out a similar study but looking at individual performances within those sides to try and identify which players were able to step up quickly and, of course, those who did not or could not. The factors that were discovered in this study can now be used as a comparative tool to assess those individual’s performances to add a greater degree of accuracy to the raw data.