Blue Yonder’s AI solution spends most of its time analyzing vast amounts of data to predict customer demand and optimize pricing across the retail sector. However, this summer, we took on our biggest challenge yet: predicting the results of the 2018 FIFA World Cup.
Analyzing every international football match since 1872, approximately 38,000 matches, and running 1 million simulations of the World Cup enabled Blue Yonder to determine the likelihood of progression through the rounds and ultimately victory for every team.
Predicting the Unpredictable?
Despite this tournament’s status as one of the most unpredictable ever, Blue Yonder’s AI impressed with its predictions. Sweden’s victory over a Mexico side that had just beaten Germany came as a shock to all, except Blue Yonder’s AI, which predicted the result. Blue Yonder also successfully predicted the surprisingly strong performance of Russia. As the lowest FIFA-ranked nation in the competition (70th in the world), many did not see Russia even getting out of the group, let alone beating Spain to reach the quarter-finals. However, Blue Yonder’s ability to understand the nuance of home advantage saw them tipped as the 7th most successful team, with a 4.6% chance of eventual victory.
Such was the success of Blue Yonder’s forecasts that it was named second most successful predictor of the group stages by the German analytics website Bstat, above the likes of Goldman Sachs and the Technical University of Dortmund.
As the tournament advanced, it continued to earn its reputation for unpredictability, as underdogs knocked out their stronger counterparts; Russia beating Spain and Belgium triumphing over Brazil proving the biggest upsets to Blue Yonder’s predictions. Happily, for fans of the Three Lions, Blue Yonder marked England as 6th most likely to win the tournament, but Harry Kane and co. managed to exceed expectations to finish 4th, after narrowing losing in their semi-final clash against Croatia.
The ultimate winner of the World Cup 2018 was France, and despite not being Blue Yonder’s number one pick at the start, ‘Les Bleus’ were predicted as favorites going into every one of their seven games. Some teams continued to defy the odds, most notably the semi-finalists Belgium and finalists Croatia. Despite Blue Yonder predicting losses along the way and placing them as 9th and 12th most likely victors before the competition, both these teams were able to finish in the top three.
Football is Fickle, Retail isn’t Random
Football is fundamentally random. The difference between those who advance through a tournament and those who are eliminated can be a matter of inches. Off days, bad luck, crowd support, own goals and countless more variables make football inherently unpredictable, particularly as the result is so heavily dependent on the actions of just a few individuals.
However, the consistency and scale of the data used by Blue Yonder’s AI for replenishment and price optimization ensures that its predictions are significantly more accurate. Vast quantities of data on sales patterns, customer footfall, weather conditions, local events and more enables a high degree of precision when predicting customer demand or optimizing prices in a retail environment. The data allows for Blue Yonder’s AI to account for factors that influence demand and adjust prices or replenishment accordingly, delivering consistently better decisions to retailers and automating manual processes.
Even if football can be won or lost with a stroke of good fortune, when it comes to analyzing and acting upon the huge amounts of data gathered by supermarkets, ecommerce giants or fashion chains, Blue Yonder’s AI leaves little up to chance.