What is simulaciones_mundial?

This project uses data and statistical modeling to predict the results of the 2026 FIFA World Cup. It is written primarily in Spanish, and it walks through the full pipeline from collecting match data all the way to simulating the entire tournament 10,000 times to estimate each team's chances of winning. The first step is scraping historical match statistics for national teams from a sports data site called FlashScore. The scraper collects results, possession figures, shots on target, corners, fouls, goalkeeper saves, and passing numbers, plus each country's FIFA ranking. Pre-scraped data is also included in the repository if you do not want to run the scraper yourself. Next, the data is cleaned and combined into a single table with one row per match. From there, the project builds rolling averages (covering the last five matches and the full historical record) and computes differences between the two teams in each statistical category. These processed numbers are what the models actually see when making predictions. Two separate models are trained using a library called XGBoost. The first model predicts how many goals each team is likely to score, using a statistical distribution suited to football scores. The second model uses the first model's goal predictions as inputs and estimates the probability of a home win, draw, or away win, with an extra calibration step to make those probabilities more accurate. The project also includes time-aware validation to prevent the models from training on data that comes after the matches being predicted. The Monte Carlo simulation runs 10,000 complete tournaments, covering all 104 matches from the group stage through the final. Because football has a lot of randomness, running many simulations is more informative than running just one: a team with a 51 percent chance of advancing will do so in roughly 5,100 of the 10,000 simulations, while a stronger team with a 99 percent chance advances in nearly all of them. The results, including the most likely scoreline for each match and estimated champion probabilities for every country, are saved as a readable report and as CSV files. A static web page built for GitHub Pages displays daily predictions and updates automatically.