The brief.
Take a quantitative concept — "how does Bitcoin's price move historically relative to itself, and what does that say about where it could go next" — and build it from raw exchange API to interactive dashboard in a single focused engineering project. No frameworks. No SaaS dashboards. Single-file Python that runs on any laptop.
Architecture.
Data Pipeline
- Multi-timeframe data fetching (1h, 4h, 1D, 1W, 1M) from Bitfinex with intelligent backfill and incremental updates
- Local CSV cache so subsequent runs only fetch the new bars since the last sync
- Seven years of historical data, kept on the local filesystem for fast statistical analysis
Statistical Engine
- Rolling-window percentile band computation across configurable lookback periods
- Absolute and percentage-delta modes for the same underlying analysis
- Laddering calculator that takes a hypothetical entry / exit and computes position size against the statistical bands
Presentation
- Self-hosted HTTP server with order-execution endpoints (sandbox/testnet only by default)
- SQLite for historical state retention across runs
- Single-file Python implementation, no framework dependencies — runs on any laptop that has Python
- Tabbed interface with chart, calculator, liquidations heatmap, and integrated YouTube feed of relevant macro commentary
What it demonstrates: We can take a quantitative concept from raw API to interactive tool in one focused engineering project. The same patterns apply directly to any domain — supply-chain analytics, manufacturing OEE dashboards, healthcare utilization analytics, financial performance reporting. When the off-the-shelf BI tools don't fit, we can write the analysis tool the client actually needs.