The Trouble With Counting Stats

      Comments Off on The Trouble With Counting Stats

Something got under my skin earlier this week, and then boiled over today when I scratched beneath the surface a little, so I wanted to get a little gripe off my chest here, and hopefully we can all agree to start using counting stats a little more responsibly.

The following two tweets popped into the BoroBrain timeline this week:

Thankfully the Football League were helpful enough to share their data to save a little bit of grunt work, even if it was in PDF format. I’ve converted it into an excel document that anyone can fiddle with, and also appended some more relevent statistics, the link for which is at the foot of this article.

Ordinarily this shouldn’t warrant much attention, but the Football League’s twitter account has approaching a quarter of a million followers, so these figures will have gained some attention and indeed the club have (quite rightly) jumped all over it for a short burst of propaganda. The stats shown merely rank clubs by the number of goals they scored at home in the calendar year. This takes no account of a number of pretty important factors:

(1) division – it’s probably fair to assume that it’s harder to score goals against Championship opponents than against foes in League Two. Comparing Mansfield with Middlesborough in the same table simply isn’t justifiable.

(2) relative strength of schedule – by taking the second half of one season and the first half of another, some sides are naturally going to have faced tougher opposition in home games than others from the same division. Worse, they might well have faced the best sides twice and not met the worst sides at all.

(3) number of games played – in 2014, Boro’ were one of six clubs to play as many as 25 Football League games at home. A greater number played only 21 of them. Obviously if you have four fewer chances (that’s 16% less playing time) then you’re less likely to score as many goals in total.

This isn’t an attempt to belittle Boro’s achievement in having a stronger home record in 2014 than most, more to put it into context. To put Boro’ 7th in a list of most goals scored at home is factually accurate but also meaningless if you genuinely want to take it as a measure of performance. This why counting statistics are so dangerous if not used with care – a more meaningful way of comparing performance by club would, for example, be to eradicate that third factor and calculate the number of goals scored per game.

So, for the record, a brief summary of some crude Boro’-related 2014 statistics and then their contextual equivalents (rankings displayed out of 67 qualifying clubs to have played in the Football League throughout the year, where highest number = 1st rank):

Goals scored at home – 41 (rank: 7th)
Goals scored per home game – 1.64 (rank: 14th)

Goals conceded at home – 30 (rank: 21st)
Goals conceded per home game – 1.20 (rank: 27th)

Points earned at home – 43 (rank: 9th)
Points earned at home as a %age of those available – 57.3% (rank: 19th)

Goals scored away – 16 (rank: 64th)
Goals scored per away game – 0.76 (rank: 63rd)

Goals conceded away – 31 (rank: 33rd)
Goals conceded per away game – 1.48 (rank: 22nd)

Points earned away – 14 (rank: 64th)
Points earned away as a %age of those available – 22.2% (rank: 63rd)

In each case above, the second stat at least tries to help compare apples with apples, but ultimately it’s a pretty pointless endeavour. Football isn’t played in calendar years and the league is just trying to have some harmless fun by playing around with numbers on the last day of term, but it frustrates so much when they then insist on publishing them in this way.

I have to say, though, that seeing only 16 league goals away from home in a full year took me by surprise a little, I hadn’t appreciated that things were that significantly split between home and away. Maybe I’ll take a look at that, soon…

DOWNLOAD: Click here for the full table of club-by-club stats for 2014.