Football World Cup 2010: The winner will be...

What are the chances?  Your teams prospects in 2010 - an interesting, probability based approach

As I write this, I am sitting at the Cape Town International airport, about to fly to Johannesburg for a BBC Media launch of their World Cup 2010 coverage, now only one day away.  I'm on a panel of experts which is supposed to discuss the upcoming tournament and launch a BBC show covering the action.  I have no idea what this involves, but if anything interesting happens, I'll let you know!

All around the airport, fans wearing Mexican, French, and of course South African football jerseys are milling about, in transit before their teams' matches tomorrow.  Every few minutes,  I hear the unofficial anthem of the World Cup, the vuvuzela, being blown.  If you have never heard of a vuvuzela, don't worry, you'll soon hear plenty (perhaps more than want to)!  It's pretty clear that the fans are now out in full force in South Africa.

Predicting the result

So as we start this science of football series, I thought it would be appropriate to look first at what the fan actually cares about - the result.  I'm biased, and so my enjoyment in sport comes in part from uncovering the 'hidden side' and actually seeking to understand why things happen (I guess this makes me somewhat of a nerd...), but for most, the team, the passion and the result is what drives sport.

And of course, everyone has an opinion.  And I'd love to hear yours.  Do you think that Brazil's omission of Ronaldinho hurts their chances?  Do the Ivory Coast and Germany have a chance without Drogba and Ballack?  All feelings are welcome.

But, in the interests of analysis, let's have a look at how one might predict the results more 'objectively'.  And for this, we turn to one of FIFA's sponsors, Castrol.

The Castrol rankings - explaining the context

As much as it pains me to do so, I have to acknowledge that the corporate sector has provided one of the most interesting analytical tools that I've seen in sport.  To give you a short background, Castrol is a FIFA sponsor, which places it in the second-tier of FIFA's sponsorship hierarchy.  The top-tier is occupied by six companies - Coca-Cola, Adidas, Hyundai-Kia, Sony, Emirates and VISA - who are called FIFA Partners.  One level below that come the Sponsors, and Castrol is one of them, along with MacDonald's, Budweiser, MTN, Mahindra, Yingli Solar, Seara and Continental (in case you're interested, the third-tier, National Supporters, usually comprises host nation companies at a lower cost)

So first a brief marketing lesson, since this provides the important context for the rankings.  A sponsor pays (usually money, but often a combination of money and product/service) for the rights to use certain logos and trademarks of the rights holder.  So Castrol have paid FIFA for the use the logo, the images, and exclusive rights within their category to all aspects of its global tournaments between 2007 and 2014.  There's more to it than this - certain rights of association, such as hospitality packages, are given in addition to use of trademarks, but this is the principle.

However, for the sponsor, this is only part of the picture.  Once the rights of association are secured, a great deal of thought must go into what is called leveraging the sponsorship.  One generally does not  buy an expensive car and then hide it away in a garage.  They spend money to drive it and make sure it is seen!  And similarly, having paid millions of dollars for rights, the sponsor must spend more (often much more) on telling everyone that it is involved.  This is the strategic aspect of sponsorship, which is what I used to do for a SA-based sports business company.

There are a number of ways to leverage or 'activate' a sponsorship, and they depend on what the sponsor wishes to achieve.  But before I digress too much, let's return to Castrol and football.  Their sponsorship of football is at first glance a difficult one to activate.  They're an oil company, after all, and finding an obvious strategic link with football is not the easiest task in the world.

But what they have done is to create the Castrol Football website.  Basically, their proposition is that they rank teams and players according to some level of objective measurement.  There is a short video explaining how their player ranking system works at this site, and I'll leave that for you to watch if you'd like.

For the 2010 World Cup, they have developed what I think is a really impressive tool that predicts the outcome of the tournament.  It does so by tracking performance of all teams and players over a 10 year period, and calculating an attack/defence rating.  They've then simulated the tournament and worked out the probability of each team advancing to the next round.  The actual details are impossible to find (if you know, please share!), but presumably, it weighs up each fixture and using the attack vs defence rating (a good attack vs a weak defence is likely to produce a clear-cut victory, for example), and then calculates probabilities of each result. 

Even this is somewhat beside the point.  The validity of their ranking is of interest, and I'd love to know how it works, but for this post, I'm interested in what they've done only as a means for discussion.  The prediction tool is here, and you can use it as a catalyst for your own predictions if you'd like!

From a marketing point of view, it's a brilliant tool, which drives traffic to the site, which in turn drives awareness of the brand.  And the value behind the analysis is that Castrol is communicating that it is a top-level, scientific, state-of-the-art brand, and that it invests in football.

But let's not get too carried away with marketing.  Let's rather look at what the Castrol Prediction tool thinks will happen in South Africa in this World Cup.

The whole tournament - Brazil favoured to beat Spain in the final

Below is a graph of what Castrol's probability calculator throws up as the tournament draw and results from the second round onward (click to enlarge).  You can see if their analysis gives your team a chance.

Overall tournament winning probability

In terms of probability of winning the tournament, the image below shows Castrol's predictions for each of the 32 teams.  You'll see, not surprisingly given the above, that Brazil are the favourite, followed by Spain (their predicted final opponents) and England.  Fourth are The Netherlands, but they run up against Brazil in the quarterfinals, as you can see in the above chart.  This is why even though their overall probability is fourth highest, they aren't favoured to reach a semi-final.  The Castrol prediction thus evolves match by match, which is important.  You can find your own team here too.

South Africa's chances

Now for the host nation, South Africa (you can do this on the site for any team).  Below are two images.  The first shows the likely result for the Group phase, where SA plays against France, Mexico and Uruguay.  The second shows the overall probability of South Africa making each phase in the tournament. 

Overall impressions - what probability and prediction fails to acknowledge

My point with these images is not to endorse Castrol's calculator (though I have to say, it's one of the most innovative and intriguing sponsorship leveraging strategies I have seen, speaking with a commerce hat on).  Nor am I nailing my colours to the mast and saying that I think the predictions are correct.  In fact, I'd be surprised if Castrol gets more than 50% of its predictions correct!

One of the problems with their analysis is that it is too "historically-heavy".  They have taken 10 years worth of results, and I would argue that the results of matches in 2000 are basically irrelevant to matches in 2010, since no team has the same players or coaches.  In fact, I'd go so far as to say that for some teams, 2009 results are distant memories, and new players, new team dynamics, and in some cases, new coaches mean that past performance does not predict future performance very well at all. 

The other thing that Castrol's tool cannot do is respond to current events.  Injuries, confidence and momentum (for example, France look to be in mighty trouble, but two wins and they become a dangerous team) and individual ability make sport unpredictable - that's the beauty of sport, after all. 

South Africa's case - an unquantifiable advantage

In South Africa's case, the prediction of fourth is fair if you consider that we're easily the lowest ranked team in the group.  But then you look at the momentum of the team (12 international matches unbeaten), the desire of those players to extend this dream of a home World Cup a little further, and the belief they have as a result of a growing sense of national pride and their own hard work, and you start to think that perhaps, South Africa is a dangerous team to play against.  We've seen how powerful a force home ground advantage can be - at Olympic Games, previous Football World Cups (think Korea in 2002, France in 1998 - both exceeded expectations, and dare I say it, ability) and other sports.  Home teams defy probability. 

Not that this means South Africa jumps two places to qualify - here in SA, many people have convinced themselves that a tide of national support, flags and vuvuzelas will carry the team through.  I'm afraid Mexico, Uruguay and France probably haven't received that memo.  But our chances are considerably greater than history, and the Castrol Predictor, suggests.

But the tool does stimulate great debate.  I'm aware there are a few of these predictions going around (I see ESPN did one, and there are a few others working out a game by game prediction).  But the Castrol one has a brilliant site going for it.  It's interactive, the graphics are excellent, easy to understand, and their achievement in processing vast volumes of data (validity notwithstanding) is impressive.  At the risk of sending people away from The Science of Sport, I would highly recommend playing around on it for a short time! 

And then come back here, for more analysis on the science of football!  And don't forget to sign up to our Twitter account, where I will try to post updates during matches.

Your predictions are welcome!

Ross

My prediction

P.S.  Since I'm not shy on predictions, here are mine:

Brazil to win the World Cup, because they're familiar with South Africa, and have always played well here, most recently 12 months ago, in the same conditions they'll encounter now.  They have a group of players which has been together for a long time and demonstrated the ability to win tournaments (which is quite different from winning single matches).  They also have players who have not necessarily been regular starters for their club teams.  I believe this is an advantage because at the end of a 9-month club season, fatigue is a major factor.  We've already seen a spate of injuries, and it's no co-incidence that the players getting injured are those who play a lot of football (Robben, Drogba, Essien, Rooney towards the end of the EPL).  Injuries, plus fatigue of even 5% is enough to prevent a great team from beating a good one.  So while it may seem odd, I believe freshness is an important strength.

Spain are the other team that has a great chance - their problem is the burden on key players, not only for Spain, but their clubs.  Xavi, Villa, Puyol and co are so "burdened" at their clubs that their biggest problem will be staying sharp for the whole tournament.  In their favour is that two key players, Fernando Torres and Cesc Fabregas, both missed a lot of play through injury, and so they may be sharper.  A quality side, but fall just short.

Then Germany will be good, difficult to beat as always.  A potential semi-finalist, if Diego Maradona's Argentina implode under his management.  Otherwise, if Messi finds the same linkage as he enjoys with Xavi at Barcelona, they're my pick for third.  How Maradona goes is anyone's guess - I'm not even going to try!  England's draw is relatively easy, so they look favoured for a semi-final.  But I don't think they will get there.  Look for a surprise on that side of the draw, possibly even Mexico or Serbia.

Chances are, I'm wrong!

Ross