StockSmart: Predictive Analytics for Informed Financial Decisions

Abstract

The financial market presents a variety of challenges for small and medium investors due to asymmetric information and the lack of reliable analytics tools that are affordable. This project sets its objectives on solving these challenges. StockSmart is a financial support application which provides assistance in the form of providing real time periodic analytics data and interactive dashboard to help beginner investors make informed stocks trading decisions through transparent and logical solutions. The system was developed using the Design Science Research (DSR) and a three tiers architecture, using Node.js/Express backend, a frontend using React with TypeScript and PostgreSQL for a robust database. An automatic Extract, Transform, Load (ETL) data pipeline was built using cron job scheduler in order to collect and process stocks market data from Alpha Vantage API freemium service. The uniqueness of StockSmart is the multi models prediction mechanism based on the combined three time-series models such as Simple Moving Average (SMA), Exponential Moving Average (EMA), and ARIMA. The stocks data and prediction insights are updated in real time through the utilization of a WebSocket server. The end result and back testing provided an accuracy estimates of 61% to 65% during the first phase. Overall, StockSmart achieved all of its expected objectives including an intuitive dashboard and transparent insight prediction, contributes to the current finance investing market and further reduce risks in making informed trading decisions for small and medium stocks investors.