Banks and companies try their hand at earnings forecasts.
(Photo: AFP)
Düsseldorf Dekabank is doing it, Commerzbank, UBS, Goldman Sachs have done it, and Deloitte is also there. It’s about the soccer World Cup and predicting the results of the matches. It’s about the most objective and data-supported forecasts possible as to who wins and how much.
Banks and consulting firms now like to use football tournaments to present the performance of their computing models from the disciplines of “Big Data” and “Artificial Intelligence”. Same for the 2018 World Cup.
It is more about making statistical models and algorithms tangible to the outside than about producing serious data. Deka economist Holger Bahr puts the seriousness of the study into perspective as “equally analytically clean and tongue-in-cheek,” despite the “qualitative approach”.
In the foreword of its study on the “Economics of the 2018 World Cup”, Commerzbank refers to the high error rate in the forecast. The major Swiss bank UBS analyzes using its tools for evaluating investments, and Deloitte is setting up its own betting game.
Top jobs of the day
Find the best jobs now and
be notified by email.
Central commonality: The studies probably see Germany in the final, and several times as world champion. Naturally, that goes down well in the country of the defending champion and – which had to be proven – generates a lot of media coverage.
In order to count Germany among the tournament favorites, you don’t need a statistical model, even in-depth football expertise is not absolutely necessary. The scientific approach follows the approach of using comprehensive data sets to rule out random moments as far as possible and to establish ratifiable connections – an economic approach.
So it can make sense to include factors such as home advantage, weather and opponents in the calculations, as long as they can be justified and weighted. For example, it would be possible, but meaningless, to relate Costa Rica’s away wins to the beet harvest in Münsterland. Deka, Deloitte and Commerzbank therefore use the ELO rating as a measurable variable.
The system, originally created to record the strength of chess players, is now well established in football. Broadly speaking, it takes team results into account and puts them in relation to other teams, taking into account success achieved and consistency. The details of the further algorithms remain under lock and key. These statistical methods are trade secrets – and the core of the actual business models.
The benchmark against which the economists’ tips have to be measured are the betting providers’ odds. In the long term, betting odds are based on the stakes made by the players. The initial winning odds in football are determined in a practically monopolistic manner by the company Sportradar. The Swiss data service provider records games worldwide and also monitors the betting markets for noticeable movements that could indicate manipulation.
According to the statistics, what chances does Germany have of becoming world champion again? Relatively pretty good, absolutely not quite so good. With its simulation, Deka comes up with a probability of 7.3 percent, only a victory by Brazil is higher with 7.7 percent. If the DFB selection reaches the final, it will win 60 percent, says the computer. At the same time, the chance of being eliminated in the preliminary round is 40 percent.
Commerzbank puts the probability of a German victory at 18.3 percent, which is the best value of any team. According to the model, at around 82 percent it is significantly more likely that this will not happen. Deloitte, who also include the venue in their forecasts, sees Germany against Brazil in the final. The management consultants even dare to make concrete result forecasts, which ultimately correspond to the statistically frequent result of 2:1 for the main round.
It would certainly be exciting to compare this specific forecast with that of the team at the Kickform forecasting platform. The Münster-based company also tries to take random factors into account – after all, the core business of the start-up is not investing, but recommending betting games. The formulas are developed by physics professor Andreas Heue. Accordingly, Kickform can lean further out of the window.
According to the team, this gives the team a seasonal hit rate of a good 70 percent in the top European leagues. At the World Cup, Kickform says, however, the compressed game mode causes higher deviations. But here, too, Germany has very good prospects: the probability of defending the title is estimated at 17 percent. However, Brazil is clearly ahead with 31 percent.
One thing is clear: Most of the World Cup forecasts are deliberate performance shows without guarantee. Such as at Goldman Sachs, whose earnings model for the 2014 World Cup led to the betting game embarrassment at the Handelsblatt. However, the predictions are not intended to turn football fans into betting kings, but rather to show potential customers what is technically feasible.
However, none of those involved should be sad if the tips happen to be correct.
