During the build up to our home game against Arsenal, I was browsing various forums and came across an article on an Arsenal fan website which assessed the quality of their next opponents.
It used the (now familiar to us, whether we like it or not) variable of “xG difference” — something we’re reminded of every week on Sky Sports and Match of the Day. For those who don’t know or don’t care, here’s a brief explanation so that you can at least understand for the sake of the article.
XG is basically is the
“expected goals” that a side is likely to score in a game, whereas “XG difference” is the expected difference between expected goals scored and expected goals conceded.
Indeed, during a season, there are tables out there showing which teams have the best and worst XG difference throughout the whole campaign so far — which we’ll come onto — but back to the the Arsenal forum’s assessment ahead of our home game on 8 November.
The “xG difference table” at that point in the season apparently had us ranked “slightly better than Burnley” — who were the bottom of the “XG league”. So looking forward to the game, Gunners’ fans basically had us down as a real poor side and that the game at the Stadium of Light should’ve been one of the easier three points they would get this season.
Wrong.
As we all know, Sunderland gave Arsenal one of their hardest games of the season to date in a 2-2 draw.
Furthermore, during the twelve games since then, and in what was acknowledged to be the most difficult period of our season in terms of fixtures and losing more players to the AFCON than anyone else, we produced three wins, three defeats and six draws (one of which was a cup win on penalties at Everton ) — hardly the form of a team the data lovers predict should be getting battered every week.
I’m bringing this up now because ahead of Saturday’s game against Crystal Palace, the “XG difference” stats were being produced again, this time in a lengthy article on the BBC’s own website — which you can read here.
Again the “XG difference” ranks Burnley as by far the worst team in the Premier League, with ourselves as second worst…behind Wolves.
Yes, Wolves are better than us, according to the XG purists.
The article goes on to predict that like a chain from the seabed to a ship trying to pull away, “XG difference” will win through in the end and ourselves and Aston Villa — another club not performing according to the rule book — will get dragged down the league table at some point to where “we should really be”, and if not this season, next season, when we’ll surely suffer from “second season syndrome” like so many promoted clubs before us.
Maybe as Sunderland fans we’re biased and just don’t want to believe it, but I’d say the argument that as a promoted club, we’re destined to face “second season syndrome”, which happened to the likes of Sheffield United, Huddersfield and Ipswich is — and I’m not the first to say — lazy journalism, as they’ve never looked further than the club name and where they came from (the Championship).
It ignores the fact that apart from a couple of regulars each week, this is a different Sunderland team to the one that won the Championship playoff final back in May — and that there have been way more than one or two quality recruits added to the ranks.
According to the stats, one of them (Robin Roefs) is the best-performing shot stopper in the league.
If there’s a stat that says he is, who’s to argue with it? But when watching him week in and week out, we’re not seeing somebody who’s being peppered with shots like the top prize on the shelf in a coconut shy — so what factors have been taken into account?
What’s also largely been ignored is that with the third-best defensive record in the Premier League, Sunderland have largely been set up to defend deeply and to not look to dominate possession.
For any XG proponent who thinks this is because we’re struggling to do otherwise in the Premier League, it’s worth pointing out that this is what we often did in the Championship under Régis Le Bris.
The exponents of XG will argue that in increasingly data-driven world of coaching, it’s a stat that the coaches sit up and take notice of, and that they strive to produce teams that have the best stats prior to kick off.
Maybe that’s true, but it’s not the be-all and end-all.
For one thing, if Sunderland’s XG difference is so bad, why didn’t Le Bris and his staff resign months ago?
“We’ve done all we can but the situation is hopeless. I know we’re currently ninth in the Premier League but the XG difference doesn’t lie. It predicts we’re buggered, so myself and the coaching staff have decided to step aside so that somebody else can turn the XG situation around”.
Maybe I’ve missed it but you rarely hear of managers or head coaches of any team talking of XG in their post-match interviews — except maybe when they’ve been beaten — and some of the comments on the BBC article are worth a look in order to see what the fans of all clubs make of it all.
A couple that caught my eye were along the lines of “You never hear a fan coming away from a match saying that we got beat but we won on XG” or “Manchester City had an XG of two before their cup tie against Exeter, but scored ten. How is this even possible?”
How about proposing a different set of metrics to measure how a team is doing? In it, the team that sticks the ball in the back of the net most often wins the match.
XG is a metric that myself and I suspect an awful lot of other fans hadn’t heard of ten years ago.
I have no doubt that it’s probably been used by the coaching staff of clubs up and down the country for a bit longer, but such increasing awareness of it in media analysis is fairly new. As such, is it guaranteed to be 100% accurate? Clearly the XG proponents see Sunderland, Villa and an apparently underperforming Wolves as being anomalies that’ll inevitably fall into line in order to make their data justifiable.
Alternatively, could it be that in Villa’s case, they have someone who’s recognised as one of the best coaches in world football? In our case, we know we have one of the most tactically astute coaches in English football with a bunch of quality recruits, and in Wolves’ case — despite playing well enough — they just can’t defend?
Does “XG difference” take into account any of these factors?
