TCMB Reserve Terminal
A financial data platform turning central bank series into traceable analytics and validated AI commentary.
Screens
Problem
Reserve, swap, liquidity and gold series arrive at different frequencies. Understanding a change requires its source, observation date and calculation method to stay visible together.
Solution
I built a platform that collects TCMB EVDS 3 data, stores it in PostgreSQL and presents it in a unified dashboard. Reserve changes, gold valuation effects, TL liquidity and analytical intervention signals have separate analysis layers.
Engineering approach
- Data collection: Tracks source health, series metadata, observation dates and revisions.
- Analytics: Keeps methods and limitations visible; missing observations are not presented as zero.
- AI commentary: Gemini explains a structured fact summary. Generated numbers, dates and claims pass through a validation layer.
- Product architecture: A Fastify API, Next.js interface and shared TypeScript types connect the layers.
What it shows
The dashboard helps explore reserve direction, liquidity changes and derived signals with their sources. Signals are not presented as confirmed transactions: TCMB does not publish transaction-level FX and gold operations in real time.
Source and documentation
The data model, setup, calculation methods and test documentation are available in the GitHub repository.