This paper addresses the detection of anomalous video games by analyzing disparities between key variables, including the game's price, the number of recommendations, the percentage of positive and negative reviews, and the estimated owners. The research aims to develop a sophisticated model that identifies games significantly deviating from expected patterns based on multidimensional relationships between these variables. Advanced statistical techniques were employed for outlier detection, including z-scores, which quantify deviations from the mean, interquartile ranges (IQR), which pinpoint extreme values within the data distribution, and Mahalanobis distance, which allows for anomaly detection by incorporating correlations between variables and multidimensional differences. Utilizing a covariance matrix, this method ensures precise identification of outliers, even in complex datasets with multiple correlated variables.

The detected anomalies were confirmed through appropriate visualization techniques, which enable the clear identification of exceptional cases and patterns that deviate from the expected distributions. These visualizations deepen the understanding of the anomalies, allowing for the formulation of critical questions regarding the credibility and balance of attributes such as price, ratings, recommendations, and ownership, as well as potential latent market anomalies.

Preliminary results suggest the existence of several notable outliers, including games with exceptionally high prices coupled with poor ratings, and games that have an extraordinarily high number of recommendations despite low average playtime. These findings point to potentially disruptive patterns in market dynamics, potentially stemming from marketing manipulation or inadequate recommendation systems that fail to reflect the true quality of the game. Identifying these anomalies lays the groundwork for further research aimed at reducing recommender bias and optimizing product rating systems, focusing on objective rather than subjective product characteristics.