Expected Goals – The single greatest statistical hoax in soccer analytics

Look, expected goals is a predictive model that predicts goals based upon previous goals scored.

NO predictive model uses the outcome generated to predict the outcome expected – a predictive model MUST use the underlying statistics generated prior to the outcome.

So when you see articles written by uneducated pundits (like those working for MLS Soccer . com) you need to treat those articles with the contempt they deserve.

Throughout the development of soccer statistics the one remaining constant has been the proliferation of finding individual soccer statistics that provide the ‘magic pill’ on just how to predict outcome.

There is only ONE constant in soccer – the TEAM that scores more goals, wins. Or, flipped the other way – the TEAM that prevents more goals scored against, wins.

EXPECTED GOALS, as a statistic, has ABSOLUTELY NOTHING to do with attaining that outcome.

But, hey… it makes “interesting” and garners “clicks” and that’s what American Soccer is all about, init?

Let’s “entertain” and support ignorance – just once I’d like to see Alexi Lalas, Taylor Twellman, John Strong, or Stu Holden offer the truth about the complete bollocks they shove down our throat about what they get paid to punt…

Our country remains, and will continue to remain a second rate soccer country until the current announcers tell people the truth about just what “shite” they are offering up.


By the way, the correlation of Expected Goals to Wins is less than .3126 – so even from a statistical point of view Expected Goals has NO Correlation whatsoever…

Best, Chris