League Focus: Questioning the Importance of the Goals Against Column

Who’s got the worst defence in the Premier League? In some ways, that’s an easy question to answer. Look at the table and you see Queen’s Park Rangers have let in 27, Leicester City 26 and Burnley 24. They lie, respectively, 17th, 20th and 19th in the table. In fact, as you’d expect, there’s a strong correlation between the goals against column and the table, but there are a couple of obvious outliers. Hull City and Tottenham Hotspur, for instance, have conceded the joint 13th most goals, yet Spurs are 10th and Hull 18th in the actual table. West Bromwich Albion have let in the joint 11th most goals but lie 16th in the actual table.

It was Sunderland’s 8-0 defeat at Southampton that got me thinking about this. Obviously the result was an embarrassment to Sunderland fans, and obviously it set their goal difference back hugely, but I was struck by the more general reaction. Again and again it was said that the players had stopped playing for their manager, which seemed a slightly weird conclusion to draw from one game, one that even at the time seemed freakish: an unfortunate first half followed by capitulation in the second. After all, goals are of diminishing importance: the fact of losing far more important than whether it’s by one or two or eight.

Since that game, Sunderland have conceded 2, 1, 1, 0, 0, 4 and 0 – eight goals in seven games, despite running into Sergio Aguero in peak form in the 4-1 defeat to Manchester City. But then, maybe the improvement isn’t such a surprise: in the seven games before the Southampton match, Sunderland let in seven, the second lowest tally in the league at that point. So in 14 games this season, Sunderland have let in 15 goals. In the other one, they let in eight.

It seems to me that, far more relevant than once letting in eight is that Sunderland have on 10 occasions let in either none or one during a game – which is more than Arsenal and the same as West Ham and Newcastle. Realistically, certainly for a team in mid-table or lower, those performances are the ones that matter: let in none or one and you’ll probably get a point or a win; let in two or more and you’ll probably lose – the third, fourth, fifth etc goals are far less relevant.

So I began to wonder whether defensive performance was best assessed by total goals conceded or by that total of games in which one or fewer goals were conceded. I then weighted that calculation, awarding three points for a clean sheet and one for one conceded. These were the findings.

So my attempts to differentiate between conceding nought and one were a failure – although a different weighting, privileging a clean sheet less may yield better results – but it does turn out that, within this limited sample size at least, the number of games in which a team has let in none or one is a better gauge of form than the total goals conceded.

Do you agree that one big defeat isn't as damaging as it may first appear? Let us know in the comments below