The study carried out by physicist and astrophysicist Juan Alejandro Valdivia and VíaCiencia He learns based on the results of each game.
The Soccer World Cup is, without a doubt, a sporting event that generates great interest. Although some teams are already established, there are no clear rules in soccer: sometimes the great candidates return home in the first round, while many others surprise. Thus, many get excited trying to guess the possible winners: bookmakers, media such as El País and even FIFA have created online platforms with projections of the World Cup. Chile, although outside of the competition, is not the exception. “Soccer is a complex system, and as such it is interesting to model it, particularly statistically,” says Juan Alejandro Valdivia, a physicist and astrophysicist from the Chilean Complexity Cluster at the University of Chile, and a soccer lover who developed a probabilistic model on the World Cup in Qatar together with VíaCiencia, AIM partner company. “One of our areas of expertise is the area of analytics and modeling. We support our clients in whatever they need in this area. With the World Cup, we decided to make this probabilistic model because it generated a lot of interest among people from all over the world”, highlights Carolina Pulido, manager of VíaCiencia.
The model developed by VíaCiencia It is probabilistic and it is learning. “The odds of winning or drawing matches between teams evolve over time as the matches are played. The odds are released taking into consideration the previous matches. With these probabilities, 100,000 championships are measured and for each team it is counted how many times they come out first, second, third, fourth, go to the quarterfinals, etc.”, explains Juan Alejandro Valdivia. “It is important that the model is probabilistic, because for the match between Argentina and Saudi Arabia the experts predicted a 100% probability that Argentina would win and the bookmakers had a return close to 20 times. Our model gave 66% for Argentina and 18% for Saudi Arabia. The model accounts for the random component of these matches and determines that this is possible”, he exemplifies. Every day the model is learning, therefore, as results are given, the model adapts its probabilities, "because more and more importance is given to the effectiveness of the team during the World Cup and the probabilities are forgotten initials ”, details the physicist. So far, the model has shown that the probabilities of the countries most likely to win the sporting event are: Brazil (14%), France (12%), England (9%). But there is more information: for example, for each game there are the probabilities of each team to win, qualify, among others. One way to evaluate the model is to use it to place bets. Using the probabilities of the model and the returns of a bookmaker, it is possible to invest maximizing the expected return. For example, in the case of the match Argentina vs Saudi Arabia our odds were 0.66 that Argentina won and the houses gave a return of 1.17, while the victory of Saudi Arabia was paid 21 times and our probability was 17%. With these numbers, despite the fact that Argentina will most likely win, the recommendation is to bet on Saudi Arabia since the expected return is higher. When carrying out this exercise with all the games played so far, if you bet a dollar per game, you would have a profit of 30.08. At the end the results for each match are shown.
Article taken from aimchile.cl
By: aim Chile