The Secret Soccer Analyst: How I Sold the Manager on Expected Goals

Over the last few years, writers like Rob Caroll and Ted Knutson have done a brilliant job connecting the theory and practice when it comes to applying your trade as an analyst within a club.

A phrase that will often appear in blogs like theirs is ‘common language’. Essentially, the analyst must adapt the way they speak about the game in order to suit the coach he or she is working for. Failing to do so has consequences. You may find that a coach will belittle you and call you a nerd, which on the surface is innocent but a worrying sign of how they value your work. Or they may go nuclear on you, unleashing a ‘proper football man’s hairdryer’ like poor old Gabriel Marcotti received when talking expected goals to the patron saint of the football old school, Craig Burley.

I think I’ve done a pretty good job in my time at using common football language to develop strong relationships, but I have slipped up on occasion. Ironically my biggest mistake mirrored Marcotti’s. I moved straight from a conversation about expected goals into a conversation with the head coach about ‘shot quality.’

Inevitably, I slipped, and the term “expected goals” came out. The coach and his assistant shared a suspicious look. You could see the cogs turning behind their eyes to figure out exactly what I was ‘up to’. The sad reality is that modern football managers live in a world where total control and knowing everything about what happens at their club is the only way they feel safe in what can be a brutal and at times toxic work environment.

The conversation wasn’t going well, and I could feel six months of hard work building trust with the coaching staff slipping away. It was time for my Hail Mary. I’d imagined this scenario on multiple occasions, and played it out in my head, but never thought it would actually come to fruition.

“Listen, we can stand here all day talking about this and probably get nowhere, or next time we have a staff game I can show you and I think you’ll be open minded enough to understand”. I’ll be honest, I didn’t plan to challenge their open mindedness and shouldn’t have, and I did shit myself a little bit as soon as the words came out. Even so, it worked perfectly. The cogs began turning behind the gaffer’s eyes once again and within 15 minutes we were out on the training pitch with a confused and intimidated academy goalkeeper between the posts.

At this point I think it’s important to note this was literally the perfect storm of opportunity, coming off a good win and catching the coach in a good mood. You should in no way try this yourself.

I asked the coach to shoot from the edge of the box. He missed. “Well then gaffer I presume you’ve never scored from outside the box in your life then” Confused, he snaps back “course I fucking have”. I take a shot and I also miss. We keep shooting for a while, doing the usual of pointing out where you’ll put it etc before I then challenge him to a game.

The rules are simple. We take ten shots each from 22 yards (the edge of the D) and the person closest to guessing the number of goals doesn’t have to make the tea. Back inside, away from the wind and cold, I personally guess three, whilst the coach guesses seven.

Twenty shots come and go, and when all is said and done, he beats me 4-2. Then we have the following conversation:

“You got lucky with a couple of them gaffer, the academy lad let you win. As soon as I went 2-0 up I realized I was 10 shots away from a free transfer.”

“Any muppet can score once or twice but my quality showed when it mattered towards the end,” he said jokingly.

“I’m surprised though, I thought you’d be better than 4/10 from there.”

“It was only ten shots though, give me a hundred and I bet I’d score more than 50.”

“So it’s almost like a small number of shots isn’t a fair way of predicting how many goals should have been scored…”

Silence.

“This is my point, and why I asked you to come out here. If you shoot ten times from here, you might score five or score none, but that doesn’t mean that if you score none you literally would never ever score from here again. If it was possible for you to take a thousand shots, wouldn’t that be a fairer way of knowing how many you would normally score?”

“Well…yeah.” He’s starting to understand.

“Okay, so using data of where every shot in every game comes from, we can look at all the shots from the last five years across the world, and from this exact spot there might have been 100,000 shots. Of them lets say 10,000 have been scored. Do you think that’s a fairer way of evaluating how often we should score?

“Yeah I guess so.”

“Okay, so when I say ‘expected goals’ this is what I mean. When we take a shot from here, using all these shots from the past we can say quite confidently that there’s a one in ten chance of this being a goal.”

The light bulb goes on.

“And if when I say after a game that we had 2.5 expected goals, this means that the shots we took would usually add up to 2.5 goals if every shot was scored at the same rate of all the ones we’ve added up in the past.”

He gets it.

“But what if that’s a volley? Volleys from here are loads harder than a normal shot.”

“That’s what’s even better boss. There’s been so many shots in the past, we even have different numbers for things like volley’s, headers, set pieces and loads more. So when two shots come from the same place, but in different situations, they are ‘worth’ different expected goals.”

“Oh….”

The conversation winds down at ths point, but the next day he asks me to put together something more formal on expected goals, and present it to him and the coaches. I create a slide show with one of Michael Caley’s shot location maps showing how many shots are typically needed from each area to score. We talk about crossing, headers, through balls and have video examples to match.

After that, things completely change between myself and the coaches. Whenever shot quality is mentioned, someone would make a joke: “Don’t you mean expected goals?” Coaches would ask me for figures post-game and I would even get the occasional “how many expected goals does player x have this year?” Even jokes about “how many expected moans will a certain coach make this week?”, which are innocent on the surface, are a sign they have got the message.

Six months later the coach moved on to another club. I was asked to come with him, but for a number of reasons I had to turn the offer down. I’ve since had calls from the analyst in place there, who couldn’t believe that he knew all about expected goals.

In a world where the coach controls everything, it is incredibly important to be able to adapt what you do in order to communicate your methods in a way that they will understand. If you don’t, you aren’t doing your job. After all, they are the one who is getting fired when the team loses, and they are the one who getting bashed on fan forums & by the media. As an analyst, you are serving the club by serving the coach and if that’s something you can’t get on board with, you’re not going to last long.

In taking the data lesson to the pitch, where the coach feels most comfortable, he’s back in total control. At a desk, with spreadsheets and graphs, the coach is vulnerable, he’s out of his comfort zone and the barriers will be up. On the pitch, shooting, he’s the alpha male. He wins the game 4-2 and proves himself superior, but at the same time I’ve won my personal Analytics World Cup. It was truly one of the most fulfilling days of my working life.