Cortana managed to correctly predict 15 of the 16 World Cup games

Microsoft’s voice assistant feature for Windows Phone, Cortana, was on a hot streak. The voice assistant had managed to accurately predict the winners of the various world cup matches, although it had recently had its streak broken when they favored Brazil over the Netherlands. However with final match that saw Germany pit their football skills against Argentina, Cortana had managed to get that right. This means that Cortana had managed to accurately predict 15 out of the 16 matches that took place over past few weeks. We’re sure Microsoft would have been happier had Cortana predicted 16 out of 16, but hey, even the experts get it wrong sometimes and 15 out of 16 is a pretty good record, considering that there are many things to consider when trying to make a prediction.

Earlier today, we asked if you thought Windows Phone digital assistant Cortana would accurately predict the final game of the World Cup with Germany playing against Argentina, and as it turns out Cortana’s prediction was correct with a German win. “Thanks to Bing, Cortana has accurately predicted the winner in 15 out of 16 games of those popular soccer matches (a perfect bracket in the knockout rounds),” Microsoft said. To help its near-accurate predictions, Cortana relied on the power of Bing to help it analyze how each team will perform. Though the World Cup pertains to soccer, it would be interesting to see how Cortana will be able to predict wins and losses in other sports. This is how the magic of Cortana works: “It looks at the strength of each team through a variety of factors such as previous wins/losses/ties in qualification matches, other international competitions and margin of the victories in these competitions. Then other adjustments are made related to other factors that might give a team advantages over others like home field advantage (for Brazil), proximity (for South American teams like finalist Argentina), playing surface (hybrid grass), game-time weather conditions, and other factors. The data can also be tuned based on win/lose/tie probabilities captured by the people wagering on the outcomes. This model doesn’t rely more heavily on any one element but rather has the advantage of dynamically and continuously adapting up until game time in an innovative way to give users the best possible prediction.”