Now there is no reason for you to know this but I have a fair knowledge of statistics and I’m starting to become alarmed by the increasing use of them in a footballing context.
Opta, Moneyball and Billy Beane Pitt have all spun a complicated tricksy web of attraction towards bamboozling data which to the untrained eye can look compelling but often, really rather often is just misleading.
Everyone understands odds, flip a coin and its 50% chance of a head (really it’s very slightly less as the coin can land on its side) but generally you get the point.
This is an experiment which we can all repeat because one coin is very similar to another coin.
Take baseball, this is one man versus another man standing in completely predictable locations using predictable equipment in a fairly controlled environment. To compare one pitcher, for example, with another pitcher is a reasonable activity. The main variables here are the comparative skill levels of the pitcher at the batter. Now a great pitcher playing against good batters could possibly show similar stats to an average pitcher playing against poor batters. So one would expect to take a sample size large enough to give a good spread of opposition so that you can see roughly what his statistics are. This is very simple version of what Billy Beane used to great effect (obviously to a whole greater level of complexity)
Let’s now take this on to football, or as they should say in America, football. What with it being played with foots n’all:
Ranking one missed goal scoring attempt to another goal scoring attempt, is simply not the same thing at all. The striker and keeper are unlikely to be in the same place. The defenders and fellow attackers are very unlikely to be in the same place. The position of the striker to the ball is unlikely to be the same. The time available to make the shot can vary; the grass may vary….. you get the picture.
All these variables come into play and show that goal scoring chances are never exactly the same. So with this assumption in place if you take a striker out of one team put him in another and guess what very often their performance (statistics) change (Just at Liverpool: Torres, Carroll, Keane etc).
The football season does not provide a sufficient quantity of similar enough situations to properly compare like with like. What sort of chances will Robin Van Persie have to work with compared to say Peter Crouch? So we shouldn’t look at Suarez and say his clear cut scoring opportunities are lower than the league average of 39%. Its illogical. Opta defines clear cut as :
A situation where a player should reasonably be expected to score usually in a one-on-one scenario or from very close range.
“reasonably”, “expected” – very subjective terms. Where is the keeper in this scenario. Where are the defender in this scenario. Where is the ball in this scenario. Come on this isn’t a scientific comparison.
Using statistics for football is a slightly interesting activity. It should have no more impact than making you go “Oh, weird” not, “Oh, lets sign Stewart Downing”
There are too many variables to make statistical conclusions based on so few practically identical situations. We must stop doing it.
You can use statistics accurately for things like penalties as the conditions are quite similar.
I see Suarez is getting some grief about not completing his opportunities (most of which HE created). Yep, he should have put some of those away. When a striker is in a barren patch we used to be told, is he still actually having chances to miss? If he is then don’t worry the goals will come. If he isn’t then worry.
Let’s not forget this.
If you're going to take the time to post something here, please make sure it's relevant and geared towards creating discussion. This isn't a place to advertise services or your blog, it's an extension of our site. Please treat it as such.