Jon Anderson wrote a piece for the Sunday Herald Sun on 25 February in which Glenn Luff of Champion Data (CD) responded to criticism of his team’s annual player rankings.

In it, he criticised AFL armchair pundits who “refuse to try and understand” the statistics used in CD’s processes. It is exactly this kind of condescending attitude that means that CD will never understand that football cannot be boiled down to a set of numbers to determine the ability or value of a player.

The boffins at CD would have you believe the rankings they compile are inherently authoritative because they are backed up by cold, hard, unfeeling numbers.

These are numbers that are simply reflections of reality, they would conclude, which makes them irrefutable. But a book I read recently came up with the not-so-reflective-of-reality statistic that on average humans have one testicle.

This is not to say that the team at CD have been in any way deliberately misleading in their presentation of data, merely that whilst the numbers may add up, they are certainly not true representations of what we can see.

The issue I have with CD is that they present their data as completely objective fact. But this cannot possibly be the case.

At this point I should say that I have neither the inclination nor the money just waiting to be wasted in going to my local bookshop to pay $40 for a Prospectus, so you may choose to lump me in with Luff’s lazy and unintelligent statistic-deniers.

However, my point is not to challenge the numbers CD collect and interpret, but instead their attitude toward them.

Another book I read recently, Soccermatics by David Sumpter, attempts to illuminate the beauty of the world game by using mathematical models to show how it works, but avoids the mistake deigning to reduce the sport to mere numbers.

One passage, detailing the statisticians’ attempt to create the most accurate and objective model possible for the 2008/09 Premier League season, explains the trouble with the EA Sports Premier League Performance Index, which attempts to reduce skill to a singularly measured number, ranking the best players regardless of position.

So who do you think was the best player? Well, it was Fulham goalkeeper Mark Schwarzer. He scored 7.29 on Ian and Phil’s Match Outcome Index, coming in ahead of [Gareth] Barry in second place with 7.06 and Portsmouth’s Sol Campbell in third with 6.86. Ronaldo didn’t even make it into the top 20, nor did any player from Manchester United. In fact, the list featured only one striker, Chelsea’s Nicolas Anelka, and one midfielder, Gareth Barry. All the others were goalkeepers and defenders.

The statisticians had to amend their metrics in favour of attackers and midfielders to come up with more palatable results in player rankings, better reflecting what people subjectively thought of the value of players who score and win games.

Otherwise, they would have had to admit that the cold, hard, unfeeling numbers supported the conclusion that a 37-year-old Aussie keeper was the best player in the league.

This anecdote demonstrates the fallibility of the numbers. It is not that they are wrong, but that they can easily be changed by the statisticians, showing that they are ultimately a subjective interpretation of the facts.

None of this is new to us non-intellectuals who do not try to understand the data. A few years back, the golden statistic lauded above all was disposal efficiency. But we recognised its shortcomings in light of increased switching in the back line in order to stretch an opposition.

Thus the focus shifted to metres gained and statisticians took more notice of those who carry the ball forward and kick long. Jason Johannisen would do well to dedicate his Norm Smith Medal to these statisticians.

Then Tom Mitchell racked up a half-tonne in a losing effort against the Pies and we questioned the importance of accumulating possessions. So CD came up with a brand new, state of the art statistic: assisted metres gained. This stat showed Mitchell to be the best in the league at getting his fingernails dirty and doing the grunt work so that others can kick long and attack (a conclusion with which I largely agree).

Statistics are constantly in a state of flux. One measure can be in vogue this year, but in a season or two we might find glaring weaknesses in it, and we may wonder why we ever used it to quantify a player’s ability. It would be naïve to accept CD’s subjective interpretations of objective facts.

We will let Glenn Luff criticise those, like me, who choose to disregard the rankings compiled by those who pore over numbers longer than they ever watch the game. Woe betide anyone who seeks to rely too heavily on charts and graphs and numbers and figures rather then depending on the eye (I can’t be the only one who chortles when David King presents his hypotheses in The Lab).

I choose to keep arguing my case over drinks with my mates, attempting to persuade them to see things my way. I will use numbers, but never will I think that they are completely authoritative as arrogantly as Glenn Luff seems to do.

 

Written by Sebastian Clarke