Stat-mania may be fun, but it’s also misleading

LATELY statistics seem to be playing a greater and greater role in GAA analysis. Seems every which way you look there are breakdowns of puck-out strategies, numbered accounts of half-forward/half-back clashes, and critical evaluations of one-on-one clashes.

We’re not fingerpointing; we’re as guilty as anyone else.

The fascination with figures recalls the battle in baseball analysis some years ago, which hinged on the performance of the Oakland Athletics, a relatively poor team punching above its weight under a new manager, Billy Beane.

Beane began to win when he decided to focus on an oft-neglected statistic — how often baseball hitters get on base — rather than the traditionally accepted indicators of excellence, the ability to hit home runs and a high batting average.

Don’t get bogged down in the science: what’s interesting is that along the way Beane presided over a clash in cultures, between those members of his scouting staff poring over the stats to determine which player would make the grade based on a dispassionate evaluation of his numbers, and conservatives on the payroll who tended to favour those prospects who had the “look” of big league players.

You have to allow for differences, of course. Baseball makes a fetish of the statistic, and any nerd — er, baseball fan — who dreams up a new category is a minor celebrity. As a sport based on formal confrontations within strict parameters, comparative evaluation by analysis is perfect for America’s pastime.

It’s a little different for hurling and football which are far more fluid, but the inclination to crunch numbers won’t go away. Introducing stats is like using a ratchet; you can’t put it into reverse. Teams have statisticians, numbers don’t lie, and the corollary is that football and hurling experts who use hunches and instinct to evaluate players are outdated flat-earthers.

But numbers can be used incorrectly. For instance, the basic parameters of the GAA statistic are essentially up for grabs. If you’re looking at a corner-forward and a corner-back’s individual battle, what weight do you ascribe to the height, speed and frequency of supply from out the field? To what extent must the conditions be taken into account? What importance attaches to the alignment of each team, whether the attack plays with a two-man full-forward line or the defence withdraws a half-forward to screen the full-back line?

In fact, what determines a clash for possession in the first place? If the corner-forward stands in splendid isolation and his marker is drawn outfield to cut out an overlap, when the ball is popped over that corner-back’s head to the corner-forward, can that be a clash if the two players haven’t contested the ball?

You need only look at the Waterford-Tipperary game last Sunday week. At least four times Tipperary tried a long ball to their full-forward line which trickled out over the end line. They weren’t shots on goal, but they gave Waterford four cheap restarts. Were they wides? Technically, yes, but unless you saw the game then you wouldn’t know they were attempts to bring Eoin Kelly et al into play. You’re left with a wides tally carrying an extra four notches — more than enough to skew the perception of a team’s accuracy pretty seriously.

Statistics are here to stay: proper statistics. That means a proper way of evaluating shots, possessions won and lost, clashes and plays accurately summarised: that means a lot of work. For instance, ice hockey has invented the Gretzky assist, whereby a player is credited with the pass that sets up the assist for a score. Kieran Donaghy of Kerry could be credited with at least one in last Saturday’s clash with Longford, whereas under the old dispensation he would have been simply “involved in the build-up”.

Until someone comes up with a taxonomy for statistics, we’re less than convinced about the value of the stat. Without strict definitions, GAA statistics are incomplete and thus wrong; but if you supply enough qualification about outside conditions, the precision of naming will take away from the uniqueness of seeing, as the man said.

Put another way, the amount of information needed to give a true weighting to bare statistics is simply too unwieldy at present to be truly useful. Player X wins a kick-out — but only because Y makes a decoy run and goalkeeper Z is helped by the wind, and so on ad infinitum. If the qualifying of information continues, we could end up like the warring factions in the Oakland Athletics’ clubhouse — or the encyclopaedia makers in Jorge Luis Borges’ story “Tlön, Uqbar, Orbis Tertius”, where conspiring intellectuals, anxious to describe a world called Tlon, eventually find their imagined world has been hedged about with so much background information that it replaces the real world around them.

To think you could start all of that simply by counting puck-outs.

[e-mail]