As news broke that 18 year old Manchester United phenom Marcus Rashford would be joining the England squad for Euro 2016 in France, there were, as Ted Knutson identified on Statsbomb this past week, two schools of thought on the subject.
The first is fairly straightforward: how could Hodgson NOT take the young player, one of the most exciting England prospects in a generation(ish)?
The second comes from the analytics crowd: doubt. Not strong doubt, necessarily; there was no “What is Hodgson doing?” as far as I could tell from a cursory Twitter check. It was more, “This makes sense but, to be honest, there are some question marks over whether Rashford can continuing his scoring form over the summer and into next season for United.”
The reason for this doubt is understandable—Rashford’s conversion rate is very high, but his shot volume is not. The latter tends to trend more sustainably than the former—though ultimately elite strikers need to both make more chances and score a good number of them—so there’s a chance that Rashford is not as likely continue to score a goal-a-game into next year.
But as Knutson points out, this alone isn’t a good reason to prematurely dismiss Rashford as another Januzaj or—god forbid—Federico Macheda. Interestingly, to argue why, Ted “Goalpost Leaner” Knutson reverts to language more fitting for a grizzled scout than an analytics wonk:
Having seen Rashford score two goals in living colour (I was at Midtjylland Europa League knockout tie at Old Trafford), I can tell you he looks like a genuine football player. He’s reasonably tall, has a good frame, but he’s an 18-year-old who is still filling out. He’s got good pace and thus far seems to have a great knack of timing his runs to be in the right place at the right time. That’s an incredibly valuable skill if it can be done consistently.
But Knutson adds this, which is interesting: “The very fact [Rashford] is being selected to play PL matches at all at this age suggests he’s an elite talent, and the fact that he’s scoring goes a long way toward reinforcing it.”
I want to unpack this sentence a little bit. At first glance it seems to suggest that being selected to play in the first team of a Premier League club is, in itself, the mark of being ready to play in the first team of a Premier League club. This is not only tautological, but it also assigns major soothsaying abilities to the first team coaching staff at Manchester United, including the dearly departed Louis van Gaal.
But the truth about Rashford, and indeed most players when it comes to the ‘spout’ of the talent pipeline, is far more mundane. Rashford likely earned a start against FCM because he appeared ready for it, physically and technically. There was no magic signal for the heavens. At some point, exciting young players will have to get out there and play.
Thankfully, the risk paid off and Rashford performed brilliantly, sealing his place for a few matches where he continued to score. So: first a club takes a risk, and if that risk pays off, the player is rewarded with more playing time. It’s not rocket science.
What matters far more is what happens if and when Rashford’s production drops off a bit, as his output suggests it might. Incoming Man United manager Jose Mourinho, who will want to ensure his first season goes swimmingly, may be tempted to start a ‘more experienced striker’ in his stead. Maybe Rashford eventually gets loaned out to a team lower down on the Elo chain, where not only does his development suffer, but so does his service, which in turn affects his goal tally and his future prospects—a negative spiral.
At that point, the temptation will be to say Rashford “failed,” or “wasn’t good enough,” when the reality is more complicated. Despite the stories we tell after the fact, star players aren’t always destined to be stars. Sometimes talent doesn’t have the last say, and otherwise promising careers are undone by conservative coaching staff and the inability to understand how players develop over time, and that hot scoring streaks will not last forever.
The reality is that starting young players involves risk, patience, and an understanding of how form can often wax and wane while underlying ability remains constant. We don’t have flawless tools to tell the difference, but what is out there now is far better than nothing. If it can save a potentially dazzling career, it’s worthy using.
Elo and Goalkeepers
Sam Jackson is a very, very bright soccer analyst. In reading his latest post on evaluating goalkeepers, I came across this little sermon from back in February, which is probably one of the more nuanced arguments in favour of a ‘hybrid’ scouting approach.. If you read nothing else this week, you’ll have done well.
In any case, his work this past week on improving GK metrics is also very good. Elo rankings don’t always get a ton of attention, but they offer a reasonably handy tool for helping to balance individual player stats against the relative quality of their club, and this is no more important when it comes to sussing out No. 1s. Back in 2014, Gelade, whom Jackson quotes, reasoned that GKs will simply not look as good statistically if they play for weaker sides:
…goalkeepers in weaker teams do face on‑target shots that are harder to save. Failing to account for this will lead to a systematic bias; goalkeepers in weak teams will be underestimated and goalkeepers in strong teams will be overestimated.
The reason, however, that Jackson advocates for Elo here is a tad depressing, and speaks to the very real problem of a lack of public data. Most analysts would attempt to use consistent overperformance in xG vs actual goals in order to determine if a keeper has managed to consistently stop more difficult shots. Yet as Jackson writes:
The great stinking problem with xG models is the data required to create them. Shot locations, assist type, speed of attack, and even defensive pressure, are all factors that an xG model ideally includes and the collation of this information ranges between being impossibly time consuming to entirely reliant upon the purchase of expensive Opta data. As a cash-strapped student with an interest in analysing shot-stopping performance, creativity is required!
And so, Jackson instead goes for Elo scores instead, for this reason:
With Clubelo’s Elo ratings – the Elo rating system is a method for calculating relative skill levels in competitor-versus-competitor games (Wikipedia, 2016) – a respected measure of a team’s underlying quality, I figured that this could serve as a proxy for shot qualities conceded, with the logic as follows: A team with a high Elo relative to their league’s average would concede easier shots for their goalkeeper than would a team with an Elo lower than the league’s average. I am hypothesizing, then, a link between a clubs’ Elo relative to average, and a goalkeeper’s Sv%.
Jackson’s regressions reveal that yes, he’s likely onto something here.
This is great work, however, and despite the forced workaround, there are some positives when it comes to data scarcity in the soccer analytics community.
For one, it forces analysts to be resourceful, but also to play with basic data in new and experimental ways that other analytics fields might have ‘skipped over’ or even avoided with the rise of big data. This comes back to the truism it’s best to learn to walk before you run.
For another, football is not like North American sports. Leicester City are Premier League champions, a team that found itself in League One in 2008-09. This means there is a gradation of financial resources. Not every club can afford what event data providers are willing to offer, and here some inventive thinking from an eager analytics consultant can potentially prove very valuable.
Nevertheless, it’s not a good look to have analysts resort to statistical duct tape in the absence of better data, and this is why I’m still concerned about the field stagnating without a better stats site.
“The Goal is Collaboration”
This article from Stephen Shea is excellent, and echoes some of the themes I wrote about in this post from a few weeks ago. Shea’s specialty is basketball analytics—he’s a PhD doncha know—but the themes here apply to any sport. He strongly emphasizes the importance of conversation over presentation, but also that this conversation must be framed in larger questions about the direction of the team. Please excuse the bigger quote but it’s worth reprinting in full:
And while we’re on the topic, let me emphasize that communication is part of collaboration, but collaboration requires more than communication. Communication can mean that the general manager asks the analytics team to create a draft model that ranks prospects, and then the analytics team produces a clear presentation of the results.
Collaboration begins with the general manager, scouts, coaches, and analytics team in a room addressing questions like, “What are the short and long term objectives of the organization?” and “What characteristics do we want to prioritize in a prospect?” Collaboration continues with questions like, “How will we help Cheick Diallo continue to develop?” or “What would Buddy Hield’s role within the organization be this year? in 3 years?” or “What do we believe are Henry Ellenson’s defensive limitations?”
The goal of analytics is not a hostile takeover of front offices. Analytics thrive on basketball wisdom, and NBA teams are stacked with tremendous basketball minds. So, let’s collaborate!
In other words, if you introduce brilliant analytics into a process that’s focused on answering all the wrong questions, you will get crappy output.
This also, to me, speaks to the effectiveness of Leicester City’s setup, with analysts sharing the same office with the first team coaches. I can speak from experience at theScore working next to people vastly more informed than me—this is a winning formula, and so easy to put in place. It just takes some humility on the part of those higher up on the foodchain.