The Heebie Eee Gees

Halloween has been and gone. So has the first round of the FA Cup and, for the team I support, a live-on-BBC away game at a National League club that might as well have had a banana skin icon in the bottom corner of the screen.

One comfortable-ish win later and a sigh of relief as the draw for round 2 fast approached.

That was quite enough scares averted for one weekend.

But it wasn’t the only frightening moment.

Now, some of my favourite song lyrics (by Pink - Don’t Let Me Get Me) go ‘LA told me “You’ll be a pop star.” “All you have to change is everything you are.”

I was reminded of them when – over the weekend – I received a LinkedIn message informing me that I was a ‘top applicant’ for a job. Curious as to what role it might be that I was incredibly suited for, I delved deeper. I mean, you never know.

And I was perfect for it. Well, other than the four most significant skills and previous experience required for the role; which I didn’t have.

But apart from that…it was mine to lose.

I wondered if it was a case of ‘it’s not you, it’s me.’ Had I painted a completely different picture of myself online that mislead LinkedIn to believe I was more suited than I was? I checked and I don’t think I had. It isn’t outside the realms of fantasy that I might have made myself sound a bit better – but that’s hardly a crime or surprise on social media is it? I know someone who is the Patron Saint of Motherhood online, but only just about remembers her kids’ names in real life. But no, it was them; my data was actually very accurate .

The reality here was that the data they had on me didn’t - like me - do the job.. The picture they painted using it was misleading both in their marketing, and for all concerned. Imagine the company who are recruiting and went ‘stop looking, Darren here’s a top applicant’ and then read my CV.

‘Actually, don’t stop just yet.’

Which is a very long-winded way of me getting to the topic of this article – Expected Goals.

Lies, Damn Lies and Statistics

I’m not knocking XG. Well, I am I suppose, but not the concept. I get that data is important, but I wonder how much of it serves its purpose. Like so much of the wave of statistics that have flooded into football there are some that do what they should, and some that maybe don’t.

We’ve already deduced that possession isn’t all that. My team had possession stats of less than 30% after 20 minutes in one game, and a shot count of 10:0 in their favour. If we’re looking for patterns, we might conclude that the less they have the ball, the more dangerous they are; and a tactic going forward might be to simply pass the ball to the opposition.

Except that the one data column that really mattered showed a big fat zero.

It’s become a little bit of a pattern. In the three games at the end of October their (our?) overall shot count was exactly 50 (fifty). They don’t measure XG in the lower leagues, as far as I know, but that’s a lot of chances being created.

In contrast, total shots by the opposition across the three games totalled - the average age of a combat soldier - n-n-n-n-n-nineteen.

We scored one goal overall from those fifty efforts. Which is pretty terrifying in itself. The opposition scored three, culminating in two damaging league defeats and a draw that puts us in danger of an EFL Trophy exit.

League points gained? Zero. Nul points.

That happens in a cup competition and you’re out. No questions asked. Unless it's the Champions League where you can lose half your games and still win the whole thing. But that’s another article.

Of course there is merit in looking beyond results. There are patterns worth noting.

I recall a 27-shots-in-one-game 0-1 defeat where – despite falling to the foot of the table – it was obvious that the manager was ‘doing something right’ as was proved by a complete turnaround in fortunes shortly afterwards; one that eventually took the manager (not the team unfortunately) to the Premier League and also the club to Wembley for the first time ever.

Expectations

It would have been very easy to do what most clubs do and sack the manager based on purely results. It is perhaps unrealistic to expect them to make decisions on what might have happened, so credit to them on this occasion for seeing past the immediate negative noise and looking at the bigger picture. It certainly benefited the club in the long run.

Of course, one of the problems is that things rarely turn out as hoped. Never mind expected.

This Sunday, whilst my wife and I were in London, we expected our son to take the rubbish to the bins, put his plate in the dishwasher and refill the cats’ bowls. He did none of these things. His EJ (Expected Jobs) was 3.0 but his score for the day was much lower. A proper horror show.

So, apart from giving BBC journalists something else to write about, what does Expected Goals really tell us? If the XG is much higher than the goals scored or goals conceded is higher than the XG suggests, this apparently tells us that the forwards can't shoot, and the goalie is rubbish. Except we already know that on account of us having eyes. And a manager hardly needs reminding that the player he advocated in the pre-season transfer committee meeting now can’t hit a cow’s backside with a banjo.

And few things are better than football for introducing things we don’t really want or need.

Pointless

Data is – I theory - fine, but are we in danger of creating data for data’s sake?

It is, after all, mainly about comparison. The way it is presented is almost always side-by-side in an adversarial manner. Or it’s used against a club to highlight their apparent failings. For example the stat that tells us that one team has completed 856 passes while their opponents had just six. But if 852 of those passes were by defenders, to each other and in their own half, while two of the six by the opponents were defence splitting through balls, the picture painted of that match is as misleading as me being deemed suitable for a job I’ve never done or got the skills to do.

But if the numbers mean that little, what’s the point?

On the day after Tottenham’s 0.05XG was almost the headline after they lost to Chelsea, I checked the stats and found out that – for Manchester City’s game against Bournemouth on Sunday – I can discover how much each team walked, jogged and sprinted. I don’t only now know the XG, but also XG from open play, XG from set play and xA (which I don’t have a clue about and don’t want to know).

I also know how many passes, how many backward passes, forward passes, long balls and successful final third passes each team did. As well, of course, as the total distance in metres or miles covered by each team and, crucially, the total number of clearances.

All recorded somewhere by someone. But how much of it made any difference?

For the record, in comparison, Bournemouth had more possession, won more tackles, sprinted and walked further and completed more passes (521 to 493) than City.

Their XG was rubbish though. City scored more goals, which seemed to be quite important.

I think that possibly, all the data could be replaced with a line that said they had Erling Haaland up front and that tells you far more about the match than any stats.

So, the weekend had left me not only relieved and excited about the cup draw, but thinking about numbers in a whole new way.

Pink goes on in that song to say she’s sick of being compared to ‘damn Britney Spears’ which is relevant here only because – over the Halloween weekend we’ve been focussing on – my daughter and her friends dressed up in costumes of all the different versions of Ms Spears before hitting the town on Friday.

And I can’t think of many things scarier than that.