5 things to think about before starting to use data analytics at your football club
Data analytics is rightly praised as the new frontier in football. Clubs are revolutionizing the approach to the game and maximizing performance by working the numbers. Moving your club into adopting a data-driven football culture, however, can sometimes be a bumpy ride.
If you are looking to implement a more scientific approach to decision-making at your club, which I highly recommend you to do, here are 5 insights that I hope can help you before starting this very rewarding but challenging journey:
1. You must start thinking probabilistically
Football’s low scoring nature added to the immense complexity of interactions within a match makes the game’s interpretation by outsiders inherently binary.
It's impossible to absorb everything that happens in a match, so relying in binary thinking gives you simple answers for hugely difficult questions, and we all love that.
Fans don’t care much about if you should’ve won when you actually lost. The media doesn’t praise the probable champion that actually finished 3rd place. No one celebrates a goal that should’ve been scored but ended in the stands.
Performance is not evaluated based on likely scenarios. It’s win or lose. Good or bad. Success or failure.
Yet, if you want to implement a data-driven philosophy at your club, you must abandon binary thinking and embrace the field of probabilities. Data analytics is only as good as its influence on the decision making process, so if your club cannot cope with knowing that there are no certainties in football, there’s no point in having data.
In fact, if the club's decision makers can't understand that losing is OK when you have a 80% chance of winning — a ridiculously high chance of winning, yet it also means that you will lose 1 in every 5 times the match is played, data analysis can be more a liability than a resource.
If decision makers don't understand that randomness might be the sole reason why a very talented team lose matches even when performing great, or vice-versa, then using data is useless.
These changes in perspective are difficult and take a lot of time, which then takes us to my second advice:
2. Analytics in football is for the long run
Much of football decisions is focused in maximising short-term performance. New players, new managers, new staff, etc, are often signed for the impact they’ll have on team’s performance now and not later. Few clubs make strategic plans for next season, let alone 3 years. If you are in a hurry, analytics won't help you much. It may even hurt your club.
I find that one of the greatest benefits of using data to improve decision making in football is the capacity it gives to adequately read the current scenario and have a clearer picture of the future paths available.
What kind of club do you want to be? How do you want to play? What kind of players do you need to achieve that? Who is the best manager for this? What are the players resources you have available now? How can they be improved? How much more or less money you will have in the future? What is the most efficient way to grow forward?
These are the sort of questions that you need data analysis to answer, and none of them is short-term.
Yes, data can and indeed should be used to understand the patterns of play and identify strengths and weakness of your team and opponents. You can use the numbers to help improving team selection and player recruitment now. All this is fine and valuable, but they won’t impact on developing the club’s business model, nor help you create sustainable growth.
If you can’t see ahead, you will only use data to deal with the effects, not tackle the causes. Data is good to deal with today's challenges, but it's much better to help you plan for the future.
3. You don’t need to invest a lot of money, but you still need to invest some
To start using an analytical approach to decision making at your club, you don’t need much. A computer with internet connection will do. It may look too simple, but in reality not all clubs have this in the football department, specially those that are used to the more traditional/experience approach to the game. What's the use of technology when all knowledge is stored in the decision maker's brain?
There’s plenty of free tools that can help you start collecting and analysing data. If you don’t have access to much resources, free online forms, for example, can help you build structured and less-biased qualitative data analysis on different issues such as player evaluation and scouting reports.
If you can implement a simple analytical process such as this, you will improve the efficiency of your decisions and, believe me, already have an edge over the vast majority of football clubs.
There are a lot of free quantitative data from the world's biggest leagues available. FBREF, Understat, Whoscored and Sofascore are some of the most comprehensive. Even if your club is miles away from the leagues covered by these sources, their numbers can give an idea about the metrics of the best clubs and players in the world, which you can use as benchmark for designing your game model or evaluating your players or recruitment targets.
If you have some money available for data, you can sign the service of ball-event-data providers through their platforms or, in more advanced cases, get their API. Wyscout, StatsPerform, Statsbomb and inStat are the most popular, and each has its strengths and weaknesses. If you can have access to more than one, or all of them, it can help you develop more comprehensive models, as different providers collect different type of events.
To make sure you extract the most value from these services, you need to use the data to develop your own models, which should go in accordance to your club’s performance methodology, style, etc. For this, you will need to have data scientists and data analysts working with you, which is key.
More money will give you more access to data from external sources and also the possibility of collecting your own data. This is perhaps the most added value data in terms of strategic development for a club. It is done through tracking data, both GPS and video-tracking — each working with different types of information.
GPS data is limited to your team and does not track the ball. With video-tracking you get ball positioning as well as information of your opponents, but don't get heart rates and some other information GPS provides.
In short, you can start a data analytics structure by creating a spreadsheet to compile the opinion of different staff in one academy player or you can have a full service video-tracking system giving you 5 million data-points per match. Regardless of your budget, it's all about collecting information and generating insights for more reasonable and unbiased decisions.
4. If data is king, context is power
You can have millions of numbers, but they won't mean much without context. Making sure you value each data point according to the actual fact it's representing is the cornerstone of implementing data analytics at your football club. In fact, if you can't provide the appropriate context to the information you are extracting, data may actually work against you.
This is tricky because the vast majority of numerical data you have access to are raw numbers and percentages, and it's the club's job to add the context that will eventually provide value to the data you have at hand.
The quantity and success percentage of passes of teams and players, for example, are easy to find. However, passes are not the same throughout the match. A long vertical pass that puts the receiving player in condition to easily score a goal is much more valuable, risky and difficult than a back pass the defender makes to the goalkeeper under no pressure.
So, in order to properly extract the adequate information about a pass, you need to know variables such as pressure, length, direction, distance, height, speed, starting and end position, who is passing, who is receiving, and so, so much more.
There are a few metrics already available that provide contextual value of a given event, such as xG, xA, EPV, xT, etc, but even these need to be considered according to each event circumstance.
Context will give you better answers for many questions you need to ask about the game. But the main question before implementing a data-driven approach at your club is:
5. How much pressure can you take?
Analytics is a double-edged sword. It gives you better understanding of individual and collective performances and, if you are doing it right (as mentioned in bullet #1), you start seeing probabilities, not results.
At the same time, however, you stop seeing the game the same way fans, media and club executives normally see. This can detach you from public perception and opinions. You may see hope when others see despair and you may recognise dangers where others view glory.
It's all great when you are winning, but it will take you to a difficult place when your team start losing, which will eventually happen. This is perhaps the biggest challenge in an analytics-driven football club. If you win when you should be losing, you will lose when you should be winning.
Both situations can become tricky and surely bring heat to the club's management. Can you handle the pressure generated from selling your top scorer who is super-valued in the market and demanding double wages when you know he's probably not going to keep this level of performance next season? Will you hold to your manager who has all the right metrics in place to be in the top of the table but lost the last five matches by suffering low-probability goals and is leading the club closer to relegation, making fans and media desperate for his dismissal?
Given the needed time, your forward will stop scoring and your manager will start winning. Can you hold your calm despite the public opinion and make the right decisions regardless of the external pressure?
If you can, then data is for you. If you can't, why bother?