World Cup Stats Prof: I was right all along

Spain's extra time goal against Holland in the World Cup final denied our resident stats expert victory in predicting the tournament winner. Or did it? Before a ball had been kicked Dr Ian McHale, senior lecturer in statistics at Salford University, had modelled the teams' performances in the run-up and had Holland as …

[Gaz Jay]

hmm

"But perhaps the real result was predicting Uruguay as third favourites after the group stage."

Not really. Looking at the wall chat after the group stages you could predict that Uruguay had an easy run to the Semi's. It was a path that England could have taken if they had topped their group.

After the groups had finished, I predicted that Uruguay would play Brazil in the semi-final (I thought Brazil would beat Holland), and that Uruguay would get knocked out at the semi and therefore go into the 3rd place playoff. Looking at the strong teams on the other side of the draw I also predicted that Uruguay would be beaten in the playoff. I didn't need to analyse any hard data for this prediction, as a football fan I just made my prediction based on my knowledge of how good teams are.

Was is amazing is that Octopus in Germany.

[jonathanb]

Just get an octopus

Much easier to understand, and much more accurate.

[Anonymous Coward]

And...

"Was Algeria holding England to a 0-0 draw an outlier?"

No, the other team cheated. They were using international level players, and had practiced.

Definitely cheating.

[Anonymous Coward]

@And....

>"Was Algeria holding England to a 0-0 draw an outlier?"

>

>No, the other team cheated. They were using international level players, and had practiced.

>

>Definitely cheating.

.... Flanders & Swan had come to this conclusion ~50 years ago in their song "the English are best" which has as its last verse

And all the world over its always the same

They've simply no idea of playing the game

They argue with umpires and cheer when they've won

And practice beforehand which spoils all the fun

[Anonymous Coward]

Statistics need a proper sample size

When a team wins most games 1-0, as Spain did, there is no way to predict with accuracy whether they will win a particular game or not. Spain played very well, but their opponents had enough chances to score and win (i.e. Robben's 1 on 1 with the goalkeeper in the final, before extended time). All 4 games in the semis / finals, 12/16 elimination games, and 34/48 of the group games were decided by a single goal or penalty kicks, or tied. Luck plays too big a factor to choose an overall winner, the best one can do is say who won't get through the group stage (although even then, France and Italy were surprises).

With a 55% win probablity, the two teams have to play at least 11 games before a trend becomes noticeable, many more to get a statistically valid sample size (to KNOW that team A is better than team B). A sample size of a single game is much too small to be statistically significant.

Re Uruguay, statistically, they should not have even been in the semi-final, because Ghana should have gotten either the hand-balled goal in the last few seconds, or their penalty should have been successful. Uruguay just got lucky...

[JimC]

That's the trouble with Soccer

Goals are such rare events that its quite easy for the better team not to win...

The equal and opposite errors, of course, is basketball, where by and large they seem just to take turns at scoring because its so easy. It would be an interesting* (well interesting if you're a sad geek like me) exercise to establish what sort of frequency of scoring would give the most reliable indication of teams' performance.

*well, more interesting than watching the games anyway...

[yoinkster]

FYI ...

It's called "football" not "soccer", please update your definition files.

[Gianni Straniero]

I am a humourless bastard...

"... the bookies had [the Netherlands] at considerably longer odds than Spain, Brazil, Argentina and England (you wot?)."

The odds offered for England were not just an expression of the probability of their victory, but a hedge against massive losses should they do the unthinkable and win.

[heyrick]

The Octopus did better...

...with no concept of mathematics, and no fancy calculations.

But, hey, IT angle: Imagine an octopus with an abacus!

[Rogerborg]

Also the octopus squirts ink

Not verbal diarrhoea.

[Velv]

Statistics

88.6% of all statistics are made up on the spot. Pure Genius! Pure Guinness! Mmmm - what a classic advert!

[Floormeister]

Required

I actually predicted 15 of the 16 group stage games correctly (only missed Netherlands-Brazil). I am much smarter than an octopus or a statistical model.

[King Edward I]

Considering...

There were actually 64 group stage games, and none of them were Holland-Brazil, I think you may have an error somewhere in your predictions :P

Unless you mean POST group stage games of course!

[Floormeister]

The required TiTLE

Of course! But it really was not that hard (my group stage predictions were pretty bad but that stupid octopus and professor did not bother with all that).

[heyrick]

@ King Edward I

POST? Power-up Self-test? I guess could be an IT angle... :-)

[LuMan]

Well, my stats were MORE accurate

I predicted that 15 of the last 16 teams wouldn't win, but one would definitely come second.

Mine's the one with the pundit in the pocket.

[Anonymous Coward]

CyberSub

"By far the most likely outcome was that the wouldn't win”

[Cunningly Linguistic]

I predicted that I wouldn't give a shit

and fuck me I was right!

[Anonymous Coward]

88% probability thay wouldn't win

For Christ’s sake don’t let Rooney hear you,

You know what happens when his brain gets overloaded.

It took a week to clean up the mess last time

[Anonymous Coward]

88% right...

and 89% wrong. That's statistics profs for you.

[Jason Bloomberg]

Predicting by goal scoring

Predicting by goal scoring is quite flawed as it doesn't really reflect the abilities of the team. Ten 'off the bar' near misses, a perfectly good goal disallowed by a ref would lose out to an unlucky deflection own-goal.

Long term performance would only be an indication of how well a team may do, shorter term performance an indication of how the team (different to the long term team) is currently doing.

It's all well and good saying Rooney's a great player over the long term but it's how he performs whilst holidayin^W playing on the day which matters.

[Robert Carnegie]

Having recently read a Dorothy L. Sayers novel about the drugs trade,

I assumed that either the octopus was being used by a Far East gambling syndicate to send messages to the World Cup teams, all of whom would have had to be bribed of course, about who should win or not win, or, more simply, the octopus simply picks at random and the teams have all been told by the Far East gamblers to fulfil the predictions made by the octopus.

[Robert Carnegie]

More than a hundred per cent wrong

Because: his odds (initIally) on Holland winning, plus his odds on Spain not winning, add up to more than a hundred per cent.

It's not strictly mathematically valid, but it's satisfying!