Football is one of the most popular sports in Indonesia, especially when it comes to supporting the Indonesian National Team. The composition of the national team players is a frequently discussed topic among the public, particularly on social media platforms like YouTube. YouTube is one of the most popular social media platforms for expressing public opinions. Sentiment analysis can help identify and address issues based on public opinions shared on social media platforms such as YouTube.The classification method used in this study is Convolutional Neural Network. The dataset was obtained through data scraping, resulting in 3,200 data points. The labeling process was conducted manually by involving three annotators. The labeling results indicate 1,036 instances of "proportional," 1,416 of "not proportional," and 747 of "doubtful."Next, preprocessing was performed on the labeled data, followed by word weighting using TF-IDF. After that, modeling was conducted using Convolutional Neural Network, and the final step involved developing an interactive web application using Streamlit to analyze text sentiment based on the trained model. The accuracy result, comparing 80% training data and 20% testing data, achieved an accuracy of 89%. Meanwhile, the sentiment analysis results show that the "not proportional" sentiment appeared more frequently than the "proportional" and "doubtful" sentiments.