Rob Mackenzie, who has held recruitment and scouting positions at Leicester City, Tottenham Hotspur and Derby County, shares his experiences with OptaPro on how to effectively use data to support and inform decision-making within football clubs.
It is now three years ago now that I was asked if I would write a blog for OptaPro concerning my experiences of technical scouting at Leicester City. I think it’s fair to say that a lot has changed in that period. Amongst other things Germany won the World Cup (again), Portugal were crowned European champions, the England national team have had three different managers, Cristiano Ronaldo has scored 145 goals, the UK voted to leave the EU & Leicester City won the Premier League. Just a quiet few years then...
One thing that has also continued to change over the last three years is the perception of performance analysis, data analysis and data analytics within the sphere of recruitment and scouting in professional football.
The aim of this blog is to reflect on some of my experiences over the last three years in the hope that it may be useful for practitioners either currently working in the game or those aspiring to do so. I will use real life examples where possible, and draw on some of the qualities that I believe people who are making a difference in the game today possess. So even though I cannot decipher what the most important characteristics within a dataset are through the application of machine learning, or calculate the probability of RB Leipzig winning the Bundesliga in their debut season in the top flight, I hope that individuals with very different backgrounds and who are at differing stages of their careers can still find this blog of some use.
While there will always be some level of resistance to new ways of working or to different ways of addressing age old problems, it is now more common than ever for clubs to have established technical scouting/recruitment analysis departments which are making meaningful and significant contributions to their clubs. Approaches that place less emphasis on a scout’s ‘gut feel’ or his/her contacts list are to not to everyone’s liking, but the support and insight that data can provide when justifying decision-making at the elite level is now much less of a surprise. It is a credit to the ways in which analytics-based practitioners within the game have managed to position themselves, evolve and find a common ground with more subjective and traditionally oriented experts. Similarly, it is a great credit to the consultants, statisticians, amateur analysts and enthusiasts (often outside of the professional game) who have dedicated significant time, energy and expertise within their own respective forums to generate new knowledge and really push the field further forward. Their efforts have resulted in the high level of debate that now typically occurs on social media, at the OptaPro Analytics Forum and within a number of professional football clubs on a weekly basis.
I am a firm believer that you shouldn’t resist something just because you don’t possess the necessary skills yourself to maximise its potential impact. Instead, be open-minded and embrace the challenge of trying to make a real difference and utilise the complimentary (and not threatening or intimidating) skill sets of others to improve your own practice.
It is not what you say… It is how you say it
Rob delivered the keynote at the Tactical Insights event, organised by Leicester City in February 2016.
The significant enhancement in both the type and volume of data that is now available, along with the improved context that can now be provided, has allowed experts to devise quite complex and insightful statistical models that can help either predict or more accurately assess players’ performances.
It is in my experience, however, that complexity can often be associated with confusion or uncertainty. For one individual’s expert knowledge to really make a difference, they have got to be able to translate it and explain it to someone who doesn’t have the same level of understanding. If you can’t do that then unfortunately it doesn’t matter what you know in the first place.
The professional football industry is historically ruthless and the opportunities for non-ex-players to be afforded credibility by their former professional playing colleagues tend to be limited. The robustness and credibility of a statistical model or data for example, are not often the priority. Instead it is the way in which you conduct yourself, the language that you use and your humility that give you the chance to let your interpretation of the data make a difference.
Within recruitment in professional football, many roles are occupied by people who do not have a data analysis background. Instead they are often vastly experienced ex-professional players who make it their business to watch hundreds of live games in order to provide accurate recommendations to their respective club’s hierarchy. Therefore one of the biggest potential issues is often a lack of common ground. Not only may traditional scouts find data-driven new concepts difficult to comprehend, but they may also be described in a language that is potentially alien to them.
The harsh reality is that it is up to the technical scout or recruitment analyst to get their message across, or no difference will be made. If they are not successful in doing this, the outcomes of relatively complex concepts won’t be given the respect and importance they deserve, simply because not everyone will be able to share their understanding of how important they actually are.
Despite feeling that I have a decent level of understanding when it comes to statistical interpretations of football performance, I have myself experienced practitioners who have left me unsure as to what they’ve just said, or to what answer they have actually given me in relation to the question that I initially asked. Does the problem here lie with me and my lack of intellect? Or does the problem lie with the practitioner who cannot explain his or her statistical interpretation? The reality is it doesn’t matter whose ‘problem’ it is as barriers have already emerged that prevent you from applying it in your daily working life.
Another harsh reality that you have to accept when working in professional football is that key decision-makers’ time is limited. If you are unable to get your message across in a concise, succinct and easy-to-digest manner then people will stop asking you the question in the first place.
With that in mind, you have to quickly get to grips with the type of language and points of reference that are being used around you. Whether you agree or disagree with different barometers of quality that are used doesn’t matter - digest them, embrace them and use them to your advantage. When trying to get your point across be humble and accept that your insight may not be acted upon (as it is the prerogative of the manager or head coach to come to their own decisions), but do everything in your power so that it is. Clever and experienced managers and coaches will make up their own minds, but they will also consider the evidence that is put out in front of them - as long as you do it in a manner where you aren’t ‘telling’ them what to do.
Contributing to a process
I have been very fortunate in my career to date to have worked for some excellent managers who have all had an unbelievable wealth of experience within the game; both as players and as managers. They have an insight into the game that I was never fortunate enough to have had and as a result their points of reference are significantly different to mine.
They may deem certain aspects of a player’s performance to be particularly important, or they may place a certain emphasis on the social dynamics of their group. As a practitioner trying to forge a relationship with these types of people I have to be respectful of their experience and learn from it where I can. To be able to influence their decisions, you have to be able to understand that the final decision (and accountability) lies with them and that your role in this process is to try and inform and safeguard their decision-making as much as you possibly can.
By adopting an evidence-based approach to profiling players and making recommendations, you can provide transparency to your work, and let your conclusions and take-home messages do the talking as opposed to your own ego or expectations. No matter how much research, profiling or analysis you compile sometimes, things don’t quite go to plan (even when you think that it really will!), so it is important to accept this and also acknowledge this when others are making decisions. If you can build trust, speak in simple terms and demonstrate a sound understanding of what you are talking about… then you have a chance of making someone listen and potentially acting upon what you say.
Some of the best examples that I have been exposed to, in relation to people influencing others in this area, are when they use other players as the point of reference. The old saying goes that ‘a picture paints a thousand words’ and it couldn’t be more relevant than within scouting and recruitment. There are thousands of players that exist that you can now statistically profile, but what key decision-maker really wants to know about all of them or even has the time to know about all of them? If you know that a player’s statistical profile is similar to someone’s in your squad then say so… but also provide context. Here is an example of how I approached a situation in my most recent role as a director of recruitment:
RM: “You know that we like X in our squad for these five reasons?
RM: “Well…this guy is doing four of those things more frequently than him at the current time and he is five years younger. While I’m not saying that he’s the perfect player just yet as he’s only young, I think he could be a great long-term replacement for X and he’s one of the few players in Europe at his age to be performing at that level. If you get chance to have a look I’d be really interested to see what you think.”
Manager: “Ok… Well get me some video then and I’ll let you know my thoughts."
By approaching the situation like this, you can paint a picture of what playing style this player may have in your manager’s mind (without even showing any video), while also providing quite a tentative yet evidence-based rationale for why someone high up in the club should assess this player. By being clever with what you say, you can still acknowledge that the manager is the expert in the relationship here, as you want his expert opinion on the player, but you have also provided evidence for why he should be seriously considered.
There are no prizes to be won for why a certain player is eventually signed; it’s about getting to the stage where the club is trying to sign someone. If you approached this in a different manner and instead opted to quote the raw numbers of the four things that the player does better than a player in your squad currently, for example, what pictures do those numbers paint? How familiar is the person that you are speaking to with those numbers being significant or really meaning something? It is so easy for messages to be lost and no matter how you get there, in my experience, it is about trying to get your message across in a way that means something to the person that you are trying to influence. Even if it means that you don’t say it how you would like to (or maybe even think you should)… this isn’t about influencing you! Be clever and subtle with what you say and how you say it.
“Hitting the ground running & finding goalscorers…”
Earlier this summer I was confronted with a situation where a number of the coaching staff had recommended a player to the manager and he was keen on my input. The player had played in both the Premier League & Championship and had also previously played with one of the coaches, but he wasn’t heavily involved in his team’s squad at the time of asking. I was quizzed on my take:
Manager: “What do you reckon?”
RM: “Well I can see why people might like him as he has experience of both this level and the league above. He is quite tenacious and aggressive, but my big concern would surround where his career has been going recently. I know that you want someone who can hit the ground running, but he’s only played 75 minutes of competitive first team football in the last six months so I’m not sure where he would be physically…”
While ‘minutes played’ is far from complex, in this case it proved to be enough evidence to influence key-decision makers' thought processes. By acknowledging why the player may have been highlighted you are not questioning anyone’s judgment, and therefore you don’t put people on the back foot, but at the same time I was able to provide simple evidence to suggest that this player in particular may not fit the “capable of hitting the ground running” remit. That evidence alone proved sufficient as when it was reflected upon by the coaches and the manager, they reached the same conclusion.
Another scenario might be that when investigating some of the best young strikers across Europe you become aware that a (relatively unknown) player’s expected goals record, who you have identified, is similar to that of a striker that one of your club’s scouts likes in the English Championship. If you are fortunate enough to work in a club where expected goals is a fully understood and applied metric then great, but if not then it is your ability to either articulate the concept to the point of it being able to make a difference instantaneously or to paint a picture of the player’s statistical profile simply by who you compare him to and why.
As I mentioned previously I have been fortunate to work with some really talented technical scouts and recruitment analysts; namely Ben Wrigglesworth (ex-LCFC now Arsenal), Bryn Davies (ex-LCFC now Tottenham Hotspur) and Alex Fraser (ex-Brighton & Hove Albion now Tottenham Hotspur), who all excel at getting their points across by using their vast knowledge of players to do so. In this scenario it would be common for one of these guys to say:
Scout: “Who have you been looking at recently then?”
Recruitment analyst: “I’ve recently come across X who is currently playing in Germany. He’s already scored six this season and given the positions that he has found himself in you might actually have expected him to have scored a few more too, but I guess you’d be more worried if he wasn’t getting into those positions as a striker as opposed to him not always scoring when he gets in there.
"He looks statistically very similar to X in the Championship in that regard as whilst he scores enough to be of interest to us, if his finishing was better then he could have a lot more. Given you [Scout] quite like X in the Championship, and this guy in Germany is younger, and would be much cheaper given he only has 18 months on his contract to run, I think he represents a decent alternative option for us to pursue as well?”
Straight away, by drawing a direct comparison to a player who is already familiar to the department, the scout in question now has a visual impression of what they might expect from the player when they see him live. Similarly a positive brief connection with the player has also been created as he has been likened to a player that they already rate. While the reason for the player having been highlighted is purely statistical, it is cleverly masked in with other factors that also make the player sound appealing. A traditional scout may then interpret this as a potentially less threatening suggestion than one that is statistically loaded.
This type of approach may seem a bit ‘round the houses’, but it is important to acknowledge and maximise the expertise that ‘field scouts’ can and often do bring to the table. In order to do that you need to keep them on side. Moreover, their knowledge of statistical metrics can often be quite limited and therefore what seems like a maths lesson to them may not be approached with much enthusiasm and you are then off to a losing start. In my experience you have to work around the strengths and skills of your colleagues.
What are we all trying to achieve?
As a technical scout, recruitment analyst, statistician or data analytics consultant, when you start conducting varying analyses in order to highlight statistically significant players, your end goal is to influence other people's decision-making. Therefore your chances of being successful are governed almost entirely by your ability to translate what you have found and what you think about what you have found. You need to get the rest of the scouting department around you to become aware of the names of players that you have highlighted, and why they need to be aware of them. That’s where it’s your time to be creative.
How you make it their business to need to know about a specific player is entirely up to you. In my experience, the more aesthetically pleasing, subtle and empowering the message, the more chance you have of somebody acting upon it. That is in stark contrast to the concept of delivering or presenting plain, boring looking spreadsheets that are brimming with different data points; something that I have also done in my career!
Understand how the people you work with operate and what they deem to be important, or not, and use it your advantage when trying to get them to ‘bite’ on players. It is as much about understanding your dataset and why certain players should be looked at, as it is the type of people that you are trying to influence. If you are able to achieve your intended outcome, of influencing decision-making and practice through analytical insights, it doesn’t really matter how you do it… just make sure that you do.