The football pundits are talking, in the cliché-ridden way they have (although cliché-ridden is, I suppose, something of a cliché itself) about the romance of the FA Cup, in that lowly minnows (they are always minnows) can occasionally defeat the big guns (they are always big guns, despite not having much relevance to minnows). Yesterday saw a clutch of Premiership clubs bite the dust (after a while, painfully florid prose comes quite easily), most notably the humbling of Chelsea 4-2 by League One underdogs Bradford City. This is the beauty of football, as José Mourinho said yesterday, although one suspects he was inwardly gritting his teeth (again, probably difficult to grit one's teeth outwardly, unless you're blessed with dentures).
But other sports often seem more predictable, perhaps because they are won by accumulating a large number of points, so that the better team or player has a chance to shine through. Whereas football matches are often lucky to scrape up a goal or two, and even those are confounded by bizarre missed chances and questionable refereeing decisions, so that the result is, literally, in the lap of the gods. It is intriguing to wonder how far you would get by collecting a host of match statistics and feeding them into an appropriate machine learning algorithm, with a view to predicting the outcome of a game, or perhaps a whole season. It can't be any more difficult than predicting the stock market, or the weather, let alone how much I need to save in a personal pension to be able one day to retire in a reasonable degree of comfort. But I suspect these things may be beyond human understanding.
No comments:
Post a Comment