‘Statistical blindness’, and importance of numbers

Date published: Friday 17th February 2017 3:30

Before wrapping up our series, I want to revisit an example from last week, for two reasons: 1) it shows the value of Expected Goals, and 2) it shows that even stats-conscious people (in this case, me) can miss the main point.

Last week we looked at the Crystal Palace – Sunderland game, which finished 0-4, but 1.6 – 0.9 in expected goals. I noted that the xG totals were misleading, because Sunderland had scored several not-so-easy chances, then sat back and let Crystal Palace attack. What I should have added was that the xG were in one sense not misleading at all: they showed that Sunderland won big not by brilliant overall play, but by brilliant finishing. And as we’ve seen several times in this series, you can’t rely on brilliant finishing to carry you for very long.

In other words, Sunderland’s 0.9 xG tells us that as decisive as the victory was, it didn’t change one of their underlying problems: the failure to create good chances. The next week against Southampton, one of the best defensive teams in the league, they wound up with a minuscule 0.2 xG and, unsurprisingly, no goals at all.

And now to wrap up, because the counter says F365 has let me spend over 10,000 words on football stats. But to say we’ve only scratched the surface is putting it very mildly. Football stats are a burgeoning field, and literally every day there are new ideas worth reading about. New measures are developed on a regular basis, many of which we haven’t even had the time to mention here. The one thing you can be sure of with football stats is that there’ll always be something more to learn.

Even better, many of the chief practitioners show and discuss their work publicly, and are more than willing to answer questions and share their thoughts. There are a number of excellent writers in the field, who can explain things far more complex than we’ve mentioned here, and write far more entertainingly doing it. There are also several good analytics podcasts, where you can hear intelligent people kick around interesting ideas, and have fun doing so.

Where’s the best place to go for football analytics? Surprisingly, it’s Twitter. Virtually all the experts have accounts remarking on current developments, and all provide links to the longer articles with the detailed analysis that’s the meat of the subject. Here are links to some of the most prominent figures in the field, all of whom are worth reading on a daily basis:

Michael Caley,
Paul Riley,
Simon Gleave,
Ted Knutson,
James Yorke,
Ben Pugsley,
Mark Thompson,
Mike L. Goodman,
Thom Lawrence,
Ben Mayhew,
David Sumpter,
Sander,
and Dustin Ward.

There’s also a great site called Statsbomb which, although not as active as in the past, has an outstanding archive of articles. It’s the site that first got me interested in analytics.

I assume anyone who’s read this far has some interest in stats. But for all that they show us, we have to remember they don’t have all the answers. Last year was a case in point: the stats kept saying that Leicester would fade. Around the middle of last season, having watched all the games to that point, and looked closely at the stats, I decided with self-important assurance that Spurs would win the title.

But it gets funnier. At a gathering of statisticians around the same time, somebody took an informal poll of about ten attendees on who would win the title. Going by Expected Goals and other numbers, half of them picked Arsenal.

So there’s a condition which we can call “statistical blindness.” Stats people sometimes say “Trust the numbers, not your eyes,” but your eyes are there for a reason. Stats are an aid, sometimes an excellent aid, to understanding. They can tell you what to look for. But you still have to look, and you have to spot the things that stats can’t reach.

To close, thanks to all the people who had kind things to say about this series. And thanks to those too who had less than kind things, because they remind you what really counts. Football is first, last, and always about love. I loved the game for almost fifty years before I even suspected there could be such a thing as football stats, and if all the numbers disappeared tomorrow, I’d love it just as much. But for me, stats engage the mind in all sorts of fascinating ways. And when the mind and heart go together, there’s a special kind of joy.

 

Peter Goldstein

More Related Articles

Comments