The text on this page was automatically translated and hence may differ from the original. No rights can be derived from this translation.
At Dialogic, World Cup fever has been reigning for some time: in an internal World Cup pool, employees are competing for first place with their own predictions. Dialogic wouldn’t be Dialogic if these predictions weren't based on a well-founded modelling approach.
Predictions based on bookmakers
To make accurate predictions, a number of researchers first focused on solid 'desk research'. Cor-Jan, Jaap, and Arthur chose to base their predictions largely on the 'odds' that bookmakers (such as Unibet) use for betting on match outcomes. The underlying idea is that these bookmakers, in order not to lose too much money, conduct extensive research to determine the most likely outcome. The three predictions differ from each other because the 'odds' are converted into predicted scores in different ways.
Regression models and game theory
Reg based his prediction on a recently released report from Goldman Sachs. The research conducted by Goldman Sachs used data from 14,000+ international football matches between 1960 and 2014 to calculate the most likely outcomes using a regression model. Bram developed his own model based on ELO ratings and transfer amounts of participating players.
Game theory
Tommy also utilised the Goldman Sachs report, and additionally modelled which predictions, given the point system in the pool, would on average yield the highest score. Making a smart choice can still earn points even with an incorrect prediction. Furthermore, he modelled which prediction, given the public predictions of other pool participants, would offer the greatest advantage if correct.
Wisdom of the crowds
In James Surowiecki's book 'The Wisdom of Crowds', it is argued that in certain circumstances, a large group of laypeople may know more than a few experts. David chose to leverage this knowledge and was guided by the predictions of the 'crowd'. His prediction consistently matched the most chosen prediction by all participants of the website wkvoorspellen.nl.
Still going with 'instinct'?
While some employees based their predictions on serious models, there are also some who admit to filling in their predictions 'based on instinct'. This strategy also seems to be paying off: at the time of writing, Marloes is just below Bram's quantitative model with 100 points. The coming weekend will reveal which approach yields the best results!