Hidden Dynamics of Football Leagues: The Predictive Power of Partial Standings
08 Feb, 2021
5 Min Read
The use of data in elite football to inform data-driven decision making is more accessible than ever before. Using research from Dr. Alexander Bond, we take a look at how being able to predict your team’s finishing position at round 10, 20 or 30 can have significant strategic implications on both coach and player recruitment, as well as wider commercial football strategy off the field.
In May 2020, we met with Dr. Alexander Bond, Senior Lecturer and Researcher at Leeds Beckett University, to discuss research that allows clubs to utilize performance data to better inform decision makers in the boardroom, at club, league and federation levels.
Dr. Bond’ research is based around a mathematical study of football league finishing positions. Let’s take a closer look at the premise of the study:
Being able to predict your team's finishing position at Round 10, 20 or 30 can have significant strategic implications for a club in terms of making data-driven decisions around how likely it is you will finish the season in a certain position.
This allows forward planning based on proven statistical trends where a team knows there is a probable chance of either pushing higher in a league, consolidating in mid table, or preparing for life in the next division down.
The Method:
Match data from every fixture between 1995 and 2017 (over 2000 games) from the four senior English leagues was analyzed, together with random match scores generated for hypothetical leagues of 20 and 24 teams.
The predictive ability of the partial standings was evaluated by computing the transition probabilities between the standings at rounds 10, 20 and 30 and the final end-of-season standings for the 22 seasons (1995 - 2017). The impact of reordering match fixtures was also evaluated.
Over 2000 games from the four senior leagues in English football were analyzed by Dr. Bond to determine predictability of final standings at differing points of the season.
Findings:
In the English leagues, team position relative to end-of-season standing became ‘fixed’ much earlier in the season than was the case with the randomized leagues. In the Premier League, the following percentage in the variance of final standings were placed with the following accuracies:
Round 10: (76.9% accuracy)
Round 20: (87% accuracy)
Round 30: (94% accuracy)
“At round 10, roughly 77% of variance in leagues is already done by game 10, which is quite a stark and bold finding”, said Dr. Bond. “There’s always going to be anomalies, but by the time we get to Christmas, which is around 20 games, or mid season, 87% of league positions are already fixed, which then really made us question why do boards and chairmans make big decisions around managerial changes at this point of the season.”
Predicted standings after rounds 10,20 and 30. How can this data inform our decision making?
The Impact of This Data on Club Decision Making:
So what can the percentages of final league positions tell us here?
Managerial Changes, Good or Bad?
Working on the 87% accuracy of final league placings after round 20, mid-season managerial changes suggest minimal changes to league position will result, but of course there are other factors to consider that can be positively impacted by managerial change. Change of manager mid-season is an especially difficult decision for any leader to make because the likelihood of significant performance changes is low. When deciding to make this change, leaders will consider the plan for next season and beyond, including the impact on squad planning and budgets.
Conversely, some coaches have historically bucked this trend by joining a club mid-season and helped a club remain in the league, for example, in recent history: Tony Pulis - Crystal Palace (2013-14) where he improved his side from 19th to 11th, Roy Hodgson - West Bromwich Albion (2011-12), improved from 16th to 11th, and Sam Allardyce - Sunderland (2015-16), improved from 19th to 17th.
In relation to the above point, it must be noted that certain ‘intangible’ factors such as internal relationships and dynamics a new manager brings to a club are factors that can’t be fully represented by a mathematical dataset. Also, before making a managerial change, clubs take a perspective in the selection process with a clear vision around an overall goal for the team and the resources needed to achieve high performance in the context of those goals.
Investing in the Winter Transfer Window: Is It Worth It?
Again, looking at the season mid-point percentage of final league positions being 87% fixed, this indicates that investing in the winter transfer market does not always yield positive returns in terms of short term goals. It may be more strategic to make medium to long term decisions to plan ahead, rather than fix a situation that is statistically unlikely to be turned around during the remainder of the existing season.
The First 10 Games Are Crucial:
As league positions at round 20 are 87% fixed and at round 30, 94% fixed, this points to the importance of a strong start to the season over the first 10 matches. Working off the statistics from this study should tell coaches in particular that they need their players in form as quick as possible or face an increasingly unlikely percentage of progressing positively from mid-season onwards.
When applying context to this data, now consider when planning for pre-season (with the performance objective being to begin the season strongly), clubs must balance commercial and performance objectives. For example, commercial objectives (certainly at larger clubs) use pre-season as an opportunity to increase revenue through fees for playing in tournaments, traveling the world to sell sponsorships, merchandise and tickets through international exposure.
Teams that do this well tend to protect the first period of pre-season for physical conditioning and then plan travel to locations where time zone and weather differences can be adapted to quickly and offer a balance between matches, recovery and travel to minimize fatigue. They also invest in protocols to minimize travel fatigue which include nutrition, daylight exposure and sleep technology. Planning early and in collaboration between commercial and football departments is therefore critical to success.
To compare, in an ideal world, performance-based objectives work best in a controlled environment where a club can apply a tactical periodization plan which gets the players physically and mentally ready to start the season strong without the distractions.
Short Term vs. Long Term
The data around finishing positions tells us that teams should be looking long term, rather than trying to turn around short term situations post round 10 that are statistically unlikely. For example: Poor team performance by Christmas/mid-season point indicates a need to begin strategizing for the following season, rather than trying to turn around an unlikely scenario by overspending during the winter window or sacking a manager needlessly.
Conclusion:
Football leagues studied across the top four divisions in England appear to conform to mathematical laws, which constrain the league standings as the season progresses. This means that partial standings can be used to predict end-of-season league positions with reasonable accuracy to inform club decision makers in the areas of player recruitment, managerial retention, long term planning and wider commercial strategy.
