Predicting World Cup results using Bayesian Analytics


Predicting World Cup results using Bayesian Analytics

Introducing the MetaMetrics Bayesian World Cup Predictor (MBWCP™)

With World Cup fever about to grip the nation, let’s kick it off with a bit of fun…

Click here to open MetaMetrics 2014 FIFA World Cup Predictor

tom cropppedWe talk a lot at MetaMetrics about the value of Bayesian approaches in Marketing Analytics, and having just set up our World Cup sweepstake in the office (I drew Bosnia and Herzegovina), we thought there were some interesting parallels.

Bayesian Analysis is a way of blending different sources of information in order to create the most logical, stable and robust models possible. We have more and more data than ever at our fingertips but how do we make sense of it all, and what is the role of business judgement and past experience in all of this? Do we blindly put all our faith in the data?

We know that data can be used in different ways to support different recommendations, so we believe that the Meta approach is key to a balanced result (Meta meaning “beyond”).

This is why we have invested in building our Bayesian Modelling Platform (BayesIQ™) which is the first of its kind able to blend multiple internal and external data sources with management hypotheses in order to create results that are supported by the data and reflect reality. Even if some of those data sources are less robust, we can still get value from them by blending them into the mix. More forward looking companies are able to leverage this data whilst preventing poor decision making by not over relying on it.

So… how might this apply to the World Cup? Well, we have multiple data sources (betting odds, FIFA world rankings, past history at World Cups, and what about the climate? How well will Russia fare in the Brazilian heat?). In addition we also have our own personal beliefs and judgement. How do we combine all of these factors to arrive at a prediction of how the teams will fare? The answer is to take a Bayesian approach. We like Bayesian methods because this is in fact how we all make decisions. We process information in the context of our experience, judgement and other knowledge on a daily basis.

So you really believe this is England’s year. The young generation is coming of age, led by the wily old Roy Hodgson at the helm. No major injuries. At last! The question is exactly how strongly do you have to believe in that for it to override the cruel reality of the betting odds, the FIFA world ranking, the heat and 48 years of under achievement? Our predictor allows you to do exactly that.

It’s totally unprotected so if you want to delve into the workings you can (yes, we do like a good Excel formula) and we often write applications using this type of approach. For the more statistically minded, no, it’s not technically Bayesian in that it doesn’t use Markov Chain Monte Carlo to sample from a defined prior distribution. Our full BayesIQ™ platform does all of that and more. Please do ask if you would like to see it in action. We’d be delighted to show how it can improve your business decision making. We knocked up the MBWCP™ in a few hours of downtime whilst waiting for some client data this week, just think what we could do for you.

In the meantime, come on Bosnia and Herzegovina! I believe in you. But the predictor says not enough sadly…..Oh well, it’s probably right, roll on 2018.

Click here to open MetaMetrics 2014 FIFA World Cup Predictor

Need a hand? Speak to Sam:



No Comments Yet.

Leave a Reply

Your email address will not be published. Required fields are marked *