<![CDATA[This study discussed the challenges of stock price prediction, which is volatile and exhibits non-linear patterns, as well as the need for an easily accessible system to present prediction results quickly. A web-based application was developed to predict the closing prices of Indonesian non-banking stocks (ASII, TLKM, and UNVR) by utilizing daily market data that were updated periodically through an application programming interface. The integrated pre-trained models used were long short-term memory (LSTM) and convolutional neural network–long short-term memory (CNN-LSTM), which were integrated into the application for one-step-ahead (t+1) inference. The system development methodology followed the software development life cycle waterfall model. Functional testing using a black-box testing approach showed that the core features ran according to the requirements, so the application was considered suitable as a web-based medium for prediction and visualization.Keywords: Stock price prediction; Web application; Deep learning; Application programming interface; Non-banking stocksAbstrakPergerakan harga saham yang volatil dan non-linear menuntut pendekatan prediksi yang adaptif serta sistem yang mudah diakses. Penelitian ini bertujuan untuk mengembangkan suatu sistem prediksi harga penutupan saham dan menyajikannya secara cepat. Penelitian ini telah mengembangkan aplikasi berbasis web untuk prediksi harga penutupan saham emiten non-perbankan Indonesia (ASII, TLKM, dan UNVR) dengan memanfaatkan data pasar harian yang diperbarui berkala melalui application programming interface. Integrasi model terlatih yang digunakan yaitu long short-term memory (LSTM) dan convolutional neural network-long short-term memory (CNN-LSTM), yang diintegrasikan ke dalam aplikasi untuk proses inference satu langkah ke depan (t+1). Metodologi pengembangan sistem mengikuti software development life cycle model Waterfall. Sistem yang dikembangkan telah berfungsi sesuai dengan kebutuhan berdasarkan hasil black-box testing. ]]>
