Building analytics into a football club

How do you win a football match? 

Easy. Score more goals than the opponents.

How do you do that?

By consistently creating more, and better quality chances than you allow the opposition.

How do you do that?

By having more skillful, fitter, and better coached players?

It doesn’t take many layers of annoying questions to reach reasonably complex areas. How do you find and recruit more skillful players?  How do you make players fitter? How do you measure what better-coached players play like?

Traditionally the response to these questions from club owners has been to outsource the responsibility of answering them to a football manager. They are effectively given control of recruitment, fitness, and coaching of the players. If they don’t deliver results on the pitch they are replaced by somebody else who is again given full control.

There are two main issues with this as an approach.

Firstly the “Goldilocks” issue of retaining staff in football.

If the manager does too badly, they get fired and you start again.

If the manager does too well he will attract interest from bigger clubs, leave, and you start again.

The manager has to be “just right” to do well enough to keep his job but not too well that other people want him. But this probably means you don’t win anything.

The second, and main problem, is that every time the manager walks away he takes almost all the knowledge of how to run a football club out the door with him. Usually taking the core people from each department with him.

The solution to this issue is found in the Sporting Director model. An individual with the responsibility of offering continuity, so even when the manager (now the head coach) leaves the institutional knowledge remains the property of the club.

“Institutional knowledge is the combination of experiences, processes, data, expertise, values, and information possessed by company employees.”

But even with this model you have potential issues. Sporting Directors get better offers, or sacked too. Is the knowledge they have gathered distributed among staff and systems? And if the head coach doesn’t share a footballing vision with the Sporting Director then the gathered knowledge has no value.

We need to build a system that ensures that we are gathering useful information AND that this knowledge is used to help the club win games.

The first step is to ask; what is the current state of institutional knowledge in this club? 

Again this question leads to more questions.

Is what we are doing now as a club going to lead to future success?

What do we do well, and what do we do badly?

How are we going to measure the success of any changes we make?

Phase 1 – Decide on the destination

In order to build something  you need to start with a plan. 

Every club has probably produced, at some point, a strategy model called “<Insert Club Name Here> DNA” or has a dusty 5-year plan from 2005 somewhere in the building.

Very few of these plans have been followed through on. Ownership changes, budgets change, managers get sacked, plans are ripped up, fresh starts and clean slates are talked about.

The problem with detailed plans is they can’t cope with the short-term and relentless nature of football. A 5-year plan won’t work with an owner who sacks people after 5 bad results.

What can work is a set of statements backed up with a plan on how to make them into a reality.

It all starts with a game model.

A game model is a short set of clear principles about how the club wants to play football. Almost all managers and head coaches will have these types of documents, certainly, newly qualified coaches will have extensive plans and linked sessions.

Simple clear statements on the non-negotiables of how we are going to approach football matches such as detailed in Jonas Munkvold’s blog below.

https://jonasmunkvold.selz.com/blog/the-making-of-a-game-model

Or google “Game model Rene Maric” for a very detailed approach.

The point is not just that these documents should exist as they already do, but they should be developed by, and the property of, the club.

Any coach in the club should have easy access to these documents, the linked sessions, the contents should be the basis of the daily work and regular updates should be made as evidence is gathered on what works best.

Although they set out the philosophy of the club they need to be grounded in pragmatic reality.

It is also not possible to say “play like Pep’s Barcelona” to a team without Xavi, Iniesta, and Messi. 

You, therefore, need to pick a long-term model but phase in the stages to reaching that end goal. If you do want to play like Barcelona then you need to ensure each and every decision you take is one that leads you a step along the road to that destination. But you have to remember if you start in Burnley it takes a long time to reach Barcelona on foot.

Phase 2 – Build in the measures of success

Some teams start from a position of being very happy with the style of football they play already and wish to continue it.

Others want to change their style, in those phased steps discussed above.

The first thing to do from an analytics point of view is to ask; what does success look like?

By having a game model we should be able to build out a set of measurable impacts on the team’s playing style.

We take our baseline of how we play now.

We look at the playing principles we want to embed and the results we want to achieve.

What should we see changing?

Imagine a game model that envisages a team building out through a deep playmaker who will spread the play into wide areas to full-backs who will progress the ball upfield.

If the coach is working with the players every day in implementing this game plan we would hope it would start to show up in the data. So you need people either in-house or consulting analytics firms (we can help with both) who know how to choose the right data sources and pull out meaningful information.

For example we might like to look for basic stats like:

Who did the playmaker receive passes from?

Who did he pass to?

How many touches of the ball did he get per game compared to the average deep midfielder?

How much progression are the fullbacks making?

We can go into greater depth with more detailed statistics.

Looking at the team shape is he obviously the deepest midfielder and playing the most passes?

What sequences started from this deep playmaker?

Are there a good mix of passes to wide positions and more vertical passes?

What are the fullbacks doing with the ball after receiving it?

The coaches would use video to assess the patterns of play against the game model criteria. 

Is the movement of receivers good enough?

Are the right pass choices being made?

This would inform their training sessions

What sessions can we do to improve the decisions on passing?

What happens if our playmaker is manmarked or pressed?

Slowly, methodically, and scientifically we observe, record, and build up a set of knowledge that belongs to the club.

Recruitment

If we do well then we face that familiar problem of bigger and richer clubs wanting our coaches and players to work for them.

This means the system is working. Only this time when the star manager or player walks out the door you retain something. Not their great leadership qualities or ability on the pitch but at least you have a detailed framework for the replacement coach and a set of criteria of exactly what you want from a player in your game model.

Your analytics department should already have a list of coaches whose teams play in a similar way, and have researched who shares your footballing vision. The new coach should have “making a contribution to the club’s knowledge base by sharing expertise” as part of their contract. 

The process of continual improvement to the game model, and openness to new ideas, is vital.

Recruitment can be greatly aided by analysts who can create model specific player searches.

Being able to sequence data and identify similar team buildup, or find players who increase threat in the same way as your game model suggests can help to narrow the field of suitable players.

Systems should also be flexible enough to find players who show characteristics that they could play in your model, even if they don’t currently play in a similar style. The important thing is that you observe and record the outcomes of your predictions to build up the knowledge base.

Player Development

Arguably the biggest advantage an analytical approach has over any other is in player development. 

Why do we coach players?

To make them better at football individually and collectively.

So we need to build in that scientific approach from the first day a player steps through the door. What can we do with this player to them as good as they can possibly be within our model of play? You need a staff with a passion for player development but again this cannot be focussed on single individual brilliant coaches. It needs to be a sustainable club-led approach. You find passionate coaches, yes, but you give them access to your central knowledge repository and task them with adding to it. Coaches will come and go but the knowledge base will increase.

What this approach doesn’t mean is that everything is done by computers and we throw away the knowledge stored in the heads of experienced staff. Far from it, in our experience, the trained eyes of experts at the clubs are good. The reason they are in head of recruitment roles or long-serving scouts is that they know how to judge a player.  

What is important is that this expertise is captured. The knowledge cannot exist solely in their heads. If they are paid by the club it is club knowledge. It is within the technical grasp of any scout to add a simple scouting score and a note/voice note, to a scouting database. 

There are technical solutions to holding the vast volumes of information in a central place (and we can advise on these) but the key element is the culture of realising why recording this data is necessary.

And this links back to our main point and the key to the whole process.

We must treat the data we gather, and the analysis of it, seriously.

If you invest all this time, energy, and resources into game modelling, centralised knowledge bases, building analytics into the process, developing players, and yet your first team play a different style and you recruit players who don’t fit the model, then what are you doing it for?

Does this approach only work for big clubs with lots of resources?

No. Resources will impact the number of staff employed, the systems that can be purchased, and the depth of research that can be undertaken. However, we work with some very low-budget teams in “minor” leagues who have a game model with which we have helped align recruit and simple analysis processes.

The key to this approach is the owners of the club. If they understand the advantages (financial and sporting) of a consistent plan regardless of the head coach it helps align priorities throughout the club. It can be scaled up as the team resources increase.

Conclusion

Start with the model.

Align the coaching throughout the club to the chosen model – in stages if the model differs from your current approach.

Embed analytics and research.

Build up a club knowledge base.

Think long term and trust that the process will get you there.

Leave a comment

Your email address will not be published. Required fields are marked *