FINANCIAL AI / RESEARCH TOOLS
Market Cerdas
Bring a research question closer to its evidence.
A stock research prototype combining market news, technical charts, fundamentals, and AI-assisted exploration in one workspace.
- My role
- Full-stack AI development
- Status
- Public ASX research prototype · Private source
- Built with
- React · TypeScript · Express · Python · Gemini
Public research prototype · Private source
The problem
Stock research draws on information that is usually scattered across news, price charts, and company fundamentals. An answer without that context is difficult to evaluate. Market Cerdas explores how a single research workspace can connect those views and make the evidence behind an AI response easier to follow. The focus is on helping a person investigate a question, with the surrounding context available for their own judgement.
What I built
My work spans the React interface, Express APIs, AI retrieval, and Python data pipelines. The product includes a chat-first research experience, News Impact sentiment views, technical chart tooling, fundamentals grading, and watchlists. Retrieved context connects the assistant to available market snapshots rather than relying only on general model knowledge. The interfaces make it possible to move between a question, a market view, and the details relevant to that research.
The research journey
A visitor can start with a stock or a question, review news sentiment, inspect a technical chart, and compare company fundamentals. The AI assistant helps connect those sources within the research workspace. Different views serve different questions: a chart makes price history visible, while a fundamentals dashboard provides another perspective on a company. The product is a research and software demonstration; its scores and generated responses should be evaluated in that context.
Current scope
The public prototype supports ASX research. Indonesian stock prediction is an active development direction, rather than a capability claimed as finished here. The application source remains private; the public documentation repository explains the product and its broad architecture. This project has reinforced the importance of separating implemented behaviour from research plans, and keeping data freshness, available evidence, and the limits of a generated answer visible to the person using the system.
