Future community correctly predicts 87% of the questions from the Game of Thrones series finale
A few weeks ago, Game of Thrones’ run came to an end with a viewership of nearly 20 million people in the US. Among those viewers, however, there was a group of who was hardly surprised by the outcomes of the series: users of the social forecasting platform Futuur.
Days before the season finale, predictions for 61 key events related to the series were available on Futuur’s mobile app — and they were right in 87% of cases. Given HBO’s show global fame for its unexpected plot twists, how could a mobile app manage to predict its ending So accurately?
Apparently, by combining a qualified crowd of forecasters with a crowd-based approach to predictions. To attract qualified forecasters to its prediction platform, Futuur partnered with entertainment-focused YouTube channels — like Coxinha Nerd — and challenged its subscribers to take part in a competition to predict the ending to Game of Thrones.
How it worked
All users who registered on Futuur started their journey with 1000 Ooms, the app’s in-game currency. From that point on, they had to decide how to invest that capital in by purchasing shares in the outcomes they believed would come true. If their prediction was correct, they’d receive 1 Oom for each share they had purchased of that outcome. If they were wrong, they’d lose their investment.
Prices for buying shares, however, fluctuated according to demand. The more people concentrated their purchases around one of the outcomes, the more expensive it would get. If at a given moment, for example, a “Yes” share of the question “Will Bran become King of Westeros?” was available for 0.7 Oom, it meant the community estimated there was a 70% chance he would become king, and a 30% chance he wouldn’t. It was then up to the user to decide to bet on “Yes” (more likely to happen according to the crowd, but with a lower return on investment) or on “No” (more risky, but possibly more profitable).
Since shares could be bought or sold at any time, there was an incentive to stay up to date with the topics at stake, just like any shareholder would do regarding companies he’s invested in. Behind-the-scenes news, interviews with cast and crew, rumors on social media, etc, could therefore be as relevant in a prediction market as quarterly reports would be in a stock market. The expected result of this dynamic was an ongoing, up-to-date forecast that efficiently aggregated information from across diverse perspectives of well-informed fans.
Track record and future challenges
After bringing over 48,000 individual forecasts related to the Game of Thrones season (brier score of 0.13), the company’s CEO and Founder, Tom Bennett, says “it’s only logical that we extend our reach into areas beyond entertainment — like politics, sports, finance and science — offering a compelling alternative to traditional polls and expert analysis”.
If previous experiments serve as indication of future success, Futuur’s initial incursion in the field of Politics should be taken seriously. In the 2018 US Midterms Elections, for example, it matched the performance of both FiveThirtyEight and PredictIt, two of the world’s most renowned forecasting services, with a Brier score of 0.16 based on 3,700 individual predictions.
If the platform manages to attract as much of a qualified crowd of forecasters to new areas as it did to Game of Thrones, it can certainly strengthen its position as a key player in the prediction market landscape.