FI
SV
ENDigia's AI beat bookmakers and hockey experts – predicted the Finnish Ice Hockey Championship
It was no surprise to those who followed Digia's AI forecast during the Finnish Hockey League (Liiga) season that Tappara sealed the championship in a game played on Sunday, 28.4.2024. The Pelicans' silver and Kärpät's bronze also went right from the AI. The silver medalist's artificial intelligence saw already at the turn of the year and the winner of the bronze game before the quarter-finals. The league prediction illustrates how AI can help companies make better decisions and, ultimately, better results.
Software and service company Digia's artificial intelligence forecast has shown signs of strength throughout the season, and with the Liiga finals now over, it swept the table for good. The AI correctly predicted the top two teams starting in December and the top four in the quarter-final round.
There has also been some development compared to the very first forecasts. For example, AI predicted Ilves would be the gold medalist in November. Tappara and the Pelicans were then ranked two and three. After this, Tappara first rose in the forecast with silver and, from December onwards, became champion. Correspondingly, Ilves' artificial intelligence dropped to fourth place in December and fifth place in January, where it remained until the end of the season.
Ice hockey expert Petteri Sihvonen, who has competed with Digia's artificial intelligence model throughout the season, believed in the Ilves championship for a long time. After the Ilves´s games ended, Sihvonen made the Pelicans his favourite for the title.
"If I think about Ilves, I was staring a bit too much at the way of playing and its implementation compared to what the AI is looking at all the time. I thought Ilves would have got its act together, but AI did not start this emotion. In my opinion, Tappara is not as strong in terms of content as the results show, which is why I am putting the Pelicans in the role I planned for Ilves earlier - as the favourite to win," Sihvonen said in a video before the final games.
The success rate of the AI model was 76% – proving its worth already in the regular season
In addition to predicting the finale, it is interesting to look at how the model performed in the predictions throughout the season. A broader examination will shed light on the strengths and possibilities of utilising the model in other areas.
When calculating how AI succeeded in predicting games during the regular season, the success rate is 76%. That's a tough result compared to both bettors and professional sports journalists.
"Jokingly, an AI forecast would have brought the bettor hefty wins this season. Similarly, in business, the model helps to make better decisions and, ultimately, better results. Digia carried out an AI forecast of the outcomes of the Liiga season to illustrate the potential of AI as a tool for decision-making and management, and this is a fairly illustrative illustration of the commercial benefits of artificial intelligence," says Digia's Chief Technology Officer Juhana Juppo.
Even before the finals, there were numerous demonstrations of the functionality of the AI model.
- The AI correctly predicted the regular season winner and the team that came second.
- AI correctly predicted nine out of ten playoff teams, while most comparison predictions were correct for eight teams.
- For the quarterfinals, the AI correctly predicted all four teams that would reach the semi-finals, while most comparison predictions were correct for only two or three teams.
"The prediction is based on data from previous games, based on which AI calculates a more likely winner game by game. Artificial intelligence learns constantly: when there are changes in the data, AI predictions become more accurate," Juppo says.
It is essential to understand what AI predictions are based on. AI should not be trusted blindly, but when used correctly, it is a very good tool for decision-making, forecasting and management.
"AI is associated with a lot of hype and even misconceptions, although it is an everyday tool like other technologies. You have to know how to use it. As with any technology, understanding its strengths and limitations is key to its effective use. Combining machine-generated insights with human intuition often produces the best results," Juppo says.
Digia's blog (in finnish) offers more lessons on how the same model helps company decision-making and what should be considered when utilising artificial intelligence as a forecasting tool.
If you want to hear more forecast analysis, please attend the webinar (in finnish)on 7.5.2024. Sign up here.
More information about the forecast and its implementation can be read here (in finnish).
Further information:
Juhana Juppo
Chief Technology Officer, Digia Plc.
tel. +358 40 172 2859, juhana.juppo(a)digia.com
Digia is a software and service company that combines technological possibilities and human capabilities to build intelligent business, society and a sustainable future. Our mission is to ensure that our customers are at the forefront of digital evolution. There are more than 1,500 of us working at Digia and we operate globally with our international customers. Digia’s turnover in 2023 was EUR 192,1 million. The company is listed on Nasdaq Helsinki (DIGIA).