This research evaluates user sentiment in reviews of five Indonesian investment applications - Bibit, Ajaib, Bareksa, Stockbit, and Pluang - using a machine learning approach. Data was collected from Google Play Store and analyzed using supervised learning algorithms including Naïve Bayes, SVM, Random Forest, XGBoost, LightGBM, as well as a Soft Voting ensemble method with TF-IDF feature representation. Evaluation was conducted through 10-Fold Cross-Validation using accuracy, precision, recall, and F1-score metrics. The results indicate that the Soft Voting model achieved the best performance, while logistic regression and SVM also demonstrated balanced performance. Overall, all applications received more positive reviews compared to neutral or negative ones. Ajaib stood out as the application with the most positive perception, both in terms of volume and sentiment dominance, while Bareksa faced challenges due to its high proportion of negative and neutral reviews. Bibit, Pluang, and Stockbit showed relatively balanced sentiment distributions, though they still need to address frequently reported technical issues. Reviews were generally written in casual language, suggesting that most users come from non-professional backgrounds. These findings imply that developers should strengthen advantages such as ease of use, speed, and relevant features while improving technical issues. A limitation of this study is that it only covers Indonesian-language reviews from a single platform, potentially overlooking perspectives from other platforms or languages. This research contributes to fintech studies, investment app UX development, comparison of ML models for Indonesian text analysis, and provides user complaint-based policy insights.