Machine learning for IGaming

Machine Learning is today often associated with robots, self-driving cars, and computers that defeat chess champions. But, is it also possible to use Machine Learning for making games both more enjoyable and fairer for users? With this question in mind, Knightec began a Data Science project together with FSport AB from Helsingborg.

FSport AB is a company specialized in developing so-called daily fantasy sports games for sports such as football, ice hockey, and golf. To participate in a game, users create one or many teams consisting of several players from matches or tournaments. Picking a player comes to a certain price, which depends on their previous performances. The teams then receive points in real time during the matches or tournaments depending on how well their players perform, and users can directly see how their teams are doing in comparison to other users’ teams. If they are lucky and skilled enough, they can win a lot of money.

“Using Machine Learning in a future version of FSport’s autofill function, it would be possible to provide users with suitable team suggestions based on probable match results.”

FSport’s challenge is to give all users as big a chance as possible to win money. Skilled users will always have some advantages, but new users should also have a good chance of winning money and have a relatively high return on investment in the games. To accomplish this, it is important to make good predictions of the players’ performances, such that the player prices can be set at correct levels. It should be almost impossible to find hidden patterns and strategies in the prices, that can potentially be exploited by skilled users to win money consistently. This is where Knightec has been able to help FSport through their competence in Machine Learning. Using a combined regression and forecasting model, Knightec has decreased the players’ individual price prediction errors. Residual analysis was also used to make sure that the prediction errors are as randomly distributed as possible, so that they do not contain any hidden patterns.

To make it even easier for new players to enter a game it is also important to be able to assist them in selecting good teams. Using Machine Learning in a future version of FSport’s autofill function, it would be possible to provide users with suitable team suggestions based on probable match results. Another possibility is to let the users themselves suggest the match results, after which they are served with several tailor-made teams that they can choose from. In this way, inexperienced users can play and enjoy FSport’s games on the same terms as the most skilled users.

Knightec has long experience within Data Analysis, Artificial Intelligence, Machine Learning, and Computer Vision. Do not hesitate to contact us if your company has needs for our expertise within these fields.

Med mångfald skapar vi kvalitet i våra tjänster

Knightec har en organisation som består av tre verksamhetsområden, 550 medarbetare och nio kontor runt om i landet. Med vår bredd kan vi snabbt sätta ihop team för utveckling av både hård- och mjukvara, kvalitetssäkring, validering och projektledning. På så vis kan vi hjälpa våra kunder från idé till löpande produktion och utveckling av både produkter och tjänster. Våra verksamhetsområden är: