Digia's artificial intelligence predicts HIFK as the champion – AI predicted overwhelmingly last season, how will it do this time?

Will the Finnish Ice Hockey Championship finally be celebrated in Helsinki after many disappointing seasons? This is predicted by Digia's artificial intelligence model, according to which HIFK will win the championship, Tappara will take silver, Ilves will take bronze place, and Pelicans will take fourth place.

The forecast is available on Digia's website (in Finnish). The forecast is updated approximately monthly.

"This year, HIFK hasn't generally been predicted to be champions like last year – but HIFK's games have been strong, and now even the AI is starting to believe in HIFK. For Pelicans and Kärpät, the forecast shows a downward spiral. Data is starting to accumulate, and artificial intelligence is gaining momentum,"  says Digia's Chief Technology Officer, Juhana Juppo.

AI is challenged by numerous changes this season

Last season, there were several demonstrations of the functionality of the AI model:
 

• The AI correctly predicted the regular season winner and runner-up team — Tappara and Ilves — in the first forecast of the season and remained consistent throughout the season.
• Before the start of the season, AI correctly predicted nine out of ten playoff teams. Artificial intelligence was the only party that predicted Jukurit for the playoffs.
• For the quarterfinals, the AI correctly predicted all four teams that would reach the semi-finals. The AI succeeded in predicting, for example, Kalpa and Pelicans reaching the semifinals before the quarterfinals.
• In the medal games, the AI correctly predicted the order of the top four - Tappara, Pelicans, Kärpät, Kalpa - even before the quarter-finals, when eight teams were left.

AI is learning constantly. So, can we expect even tougher results this season than last year? Is AI already superior?

"We must remember that the number of teams in Liiga has increased by one this season, and there is no data from Kiekko-Espoo from last season. In other words, Kiekko-Espoo is a spanner in the works for artificial intelligence," Juppo says.

The Kiekko-Espoo U-shaped curve stands out from other teams in the forecast: initially due to a lack of data, the model has not been able to predict the team very high, and as the data has accumulated, the forecast has started to improve.

"There have also been changes to other teams, coaching, players, and the league format. So there is plenty to chew on with artificial intelligence, and there is still tension in the stands," Juppo promises.

Backed by a forecasting model that also helps business leaders in decision-making

The prediction is based on mathematical modeling, in which last year's games have been "played" again with the teams' current rosters. Also, every match this season complements the model.

The forecast is based on the so-called random forest model, which combines several decision trees and makes the final decision based on the outcome of the majority of the decision trees. In a way, the AI calculates the most likely winner game by game and thus reaches the championship's final result.

"In business, the model helps to make better decisions and, ultimately, better results. The playful prediction of the outcome of the Liiga season illustrates the potential of artificial intelligence as a tool for decision-making and management," Juppo says.

AI forecast on Digia.com (in Finnish) >>

More information:

Juhana Juppo
Chief Technology Officer, Digia Plc
tel. +358 40 172 2859
juhana.juppo(a)digia.com

Digia´s communications

Photograph (HIFK-Ilves): Riku Laukkanen / @r1kuexposures