But there’s much more to say if we take this question as a starting point for some exploration of my love for football. Yes, indeed you never know what you’re gonna get in any football match. But if that were all, football would be as interesting to me as watching two people face off in a (fair) coin tossing match.
The key concept that makes football different from that is expectations. More precisely said, pre hoc expectations. Strictly spoken, all expectations are pre hoc of course, but we’ll touch on that later.
Prior to kickoff, we have our expectations about how the match is going to unfold and, most likely, which team we expect to win. The recent surge in available data has enabled us, football analysts, bloggers, pundits, analytics, fans, or whatever we like to call ourselves, to fine tune our pre game expectations with new developments almost every month. Total Shots Rate (TSR) used to be the bomb, but is now being bypassed by various Expected Goals models, using shot type, location, and what’s more.
Now, what I love about football is the perfect balance between expectations and outcomes. Roughly between fifty and sixty percent of matches see the expected winner or draw, while a slight minority of matches has a surprise outcome. Add to that the fact that football teams are generally quite unbalanced in terms of quality and we’ve got our beautiful unpredictable game that we just can’t stop wanting to predict.
This declaration of love for football is sincere, yet not unconditional. The above concept of pre game expectations and a fine rate of surprises relies on strict discipline on the expectations side of things.
All too often, football journalism falls into the trap of post hoc rationalization. And that is one mean monster, capable of eating the heart out of our beautiful, unspoiled art of matching expectations and outcomes.
Post hoc rationalization is defined here as offering a well sounding explanation for the outcome, but only after the event has taken place, generally using data, tactics and/or lines of thought to illustrate why the eventual outcome was not all that unlikely at all.
In an attempt to escape the long and dangerous arms of the post hoc rationalization monster, I’ve always tended to write my match reports in three stages. The first part, about line-up, tactics and formations before kickoff. The second part at half time, pausing the TV signal for as long as I needed, which was generally about half an hour. And the third part after the match. Though it offered a great way of (at least partly) escaping the post hoc rationalization monster, it still didn’t satisfy me and I’ve turned away from match reports on the whole, for reasons much more eloquently described here by Richard Whittall.
‘Team X has completed 70% of their tackles. Wanting it more than their opponents helped them win this match.’
Post hoc rationalization monster!
‘Playing this narrow 4-2-3-1 made them easy to defend against, because they did not stretch the defence enough.’
‘Player X completed over 90% of his passes, indicating the lack of pressure on the midfield.’
The common theme here is that causation is implied, while the only thing really observed is correlation.
‘Team X was easy to defend and they played a narrow 4-2-3-1’ does not read half as good as ‘Team X was easy to defend because they played a narrow 4-2-3-1.’
And when it reads well, it sells well.
Pundits don’t like being wrong in public, for fear of losing their generously rewarded show jobs. So, either consciously or subconsciously falling for the post hoc rationalization monster is the best way out. Your ‘analysis’ will sound, read and sell much better.
But please, play it fair, state things before the event and respect the fact that in football you’re frequently wrong!
Truth be told
Teams that win 70% of tackles, or play narrow 4-2-3-1’s, or have high pass completion rates may indeed win matches. But they may just as well lose matches, as there are a ton of other factors involved, which renders the effect of such singled out items near to nothing.
Also, teams that adjust their game in order to improve their tackling, stretch their opponent’s defense, or lower their opponent’s pass completion rate may indeed improve their chances of a favorable outcome. Or they may not, as trying to improve one aspect may just as well weaken other aspects of their game.
The present abundance of stats just leaves the door wide open for the post hoc rationalization monster to sneak in. There’s always a well sounding reasoning to build from a bunch of stats. Just line them up with the final match outcome and you’ll sound wise to the general public. Take a number that correlated with an outcome, rephrase it to cause the outcome and voilà.
But we can do better than that! Respect your readers, respect yourself. Stay away from the monster!
Love and fight
Now, back to the original question.
I love football for the fine balance between expectations and surprises, for the challenge to refine our pre game expectations with the current widely available stats, and for the challenge to stay as far away from the post hoc rationalization monster as possible!
I hope you join me in my love for the game, and in my fight against the monster!