Why perfect shouldn’t be the enemy of the good in player recruitment

Last week, the head of technical scouting at Leicester City FC Rob Mackenzie published a series of Tweets on methods clubs still use to recruit players, which have little to do with empirical assessment and a lot to do with “who you know.” James Grayson handily screencapped the comments:

 

Screen Shot 2016-02-22 at 15.02.28

While the content isn’t new, the source in this case isn’t an irate stats blogger or an angry fan but a Premier League club professional. It undermines the idea held by some analysts that because of the enormous sums involved, football clubs must employ an array of sophisticated means to assess players before agreeing to sign them for multi-year deals. For some teams, that certainly is the case. For many others however, deals worth millions of pounds in transfer fees and wages are done on the basis of passing fancy and happenstance.

One might reason the solution would be for clubs to hire a team of statisticians, technical scouts and performance analysts to completely takeover player recruitment so that expensive first team footballers aren’t signed on the basis of a handful of games, a distant memory, or a shared agent. That’s perhaps a laudable long-term goal, but for most clubs used to a particular way of doing business, it’s not realistic. It also ignores the inherent complexity of transfer business, which involves networking, negotiation, personalities, and overcoming limited time and resources.

Yet even if managers persist in signing players based solely on the most rudimentary scouting methods (“I seen him good”, to quote Grayson), an analytics-based approach can still add tremendous value to the process. To explain why, let’s consider a basic example in line with Mackenzie’s five traditional ways clubs recruit talent.   

Let’s say a manager of a top flight club nears the end of December and realizes there is a critical lack of depth in defense, particularly in the fullback department. The manager calls a meeting and sits down with their scouting department to quickly out consider some names. One is a player the head of scouting knew from a few years back who “used to be good” and might leave his current club for a reasonable fee. Another is someone the manager knows from a previous coaching job and believes he’d get on with pretty well. A third is someone the first team coach recommended to him a few weeks ago. A fourth is someone the manager knows would be easier to sign as he shares the same agent, and a fifth is one they recalled played very well against them in a pair of matches last season. The manager and scout agree on a few ballpark figures for each player, and how realistic a prospect they might be to sign during the January window.

If at this point a club were to proceed on basis of this information alone, they’d be taking on a huge amount of unnecessary risk. For example, the recruitment staff may not know that the player who played well against the club last season ranks well below the league average in sprint speed and passing accuracy, while the player with the same agent as the manager may be younger and started fewer games but has a key pass/per90 average indicative of future superstar status. Even these basic metrics would offer something more than passing, subjective “eye test.”

Even though the manager has limited his options by focusing on players whom he happens to recall, an analyst can still prevent the club from acquiring a player whom the data clearly suggests would be an expensive dud. In this case good analytics can provide some basic risk insurance, and can be a bridge between an eager manager and a nervous board in making the case to free up transfer fees on a particular player.

Ideally, teams shouldn’t limit recruitment options based solely on the passing recall of the coaching staff. A smart club would want to widen as much as possible the pool of players available to sign for a key position, in part by using accurate data to ‘find similar’ players and then apply predictive metrics and simulations that could help reveal how players might adapt. They would also want to go even further in planning player succession several seasons in advance, in order to avoid the need to panic buy at all. But even in club environments where that kind of forward thinking isn’t possible, an analyst can at the very least still assist in helping to avoid massive, multi-million pound blunders. Analytics boosters, even in their frustration at the backward thinking that persists in top flight football, shouldn’t let perfect be the enemy of the good.

About Richard Whittall

Richard Whittall has created 28 entries.

Post A Comment