I hesitated in writing this post because this could really be applied across the sporting spectrum, and, as a rule, match previews are a dying breed. Yet I think they’re worth talking about in football because they’re still considered the basis of betting in freewheeling Europe, and because the most-used predictive metrics in football are still pretty terrible. Moreover, there are some good ones around that aren’t used much at all.
But first, let’s start with the fun stuff. Match previews are, in their current form, entirely, utterly useless. The one useful characteristic of the better ones is to remind viewers which players are absent due to injury or suspension. Other than that, they’re a mere grab bag of meaningless statistics, all-time head-to-head records, likely formations (that turn out to be incorrect about one-third of the time) and in their worse iterations, mindless speculation over which will be the “Players to Watch,” or the “Head-to-Head Battles” to “Keep an Eye On.”
Even worse, because the information has no practical use, it tends to reflect the bias of the publication’s target audience. So, the Mirror‘s match preview from yesterday included these tidbits:
Schalke have won just one of their last eight games against English teams (W1 D1 L6) – a 3-1 win over Arsenal in Germany in September 2001.
Schalke have failed to score in four of their last five European games against English opponents.
All three of Aaron Ramsey’s Champions League goals for Arsenal have come in the 90th minute or later.
Arsenal’s Lukas Podolski has scored four goals in his last three Champions League matches.
Place your bets, ladies and gentleman!
Granted, match previews seem to be going the way of beta max, mostly because they’re so awful to write, and because they’re being replaced by vloggy interviews with pundits c/o betting websites and those fan blog posts that read more like secular players than anything else.
The remaining cadre of mainstream previews wouldn’t be so bad if they didn’t lead punters to ruin (or a bad night at the pub) time and time again, particularly as there are some predictive metrics out there that have slightly less of a tendency to regress to the mean. Most of them exist on obscure analytics blogs, and I’m not going to list them here, but they normally look at very broad determinations for long-term success. And few of the ones I’ve seen deal well with the Champions League.
But here’s a landmark idea you can have for free. If you’re interested in the current US election, you’ve probably visited Nate Silver’s blog, now embedded in the New York Times, fivethirtyeight.com. Silver interestingly enough got his start with PECOTA, a system to provide predictive metrics for Major League Baseball players.
Anyhoo, fivethirtyeight’s MO is essentially to statistically weigh polls (individually based on historical accuracy and sampling problems), economic conditions, and state-by-state races to produce a “win percentage” for either candidate. It’s pretty cool and it’s been fairly accurate in two election cycles in 2008 and 2010.
It would take a significant amount of work and ingenuity, but if some self-starter were to craft a similar approach—weighing various historical measures, transfer fee spending, individual player performances, etcetera—into a blog that didn’t give betting odds but instead win/loss probability by percentage, we would have no need for any match previews at all.
And that’s the genius of the thing: we’d never have to read a stupid match preview again.
So hop to it. We have nothing to lose but inane column space filler.