← back

crypto-trading-web

Python · FastAPI · React · PostgreSQL · Redis · CCXT · Docker

Can “smart money” indicators and microstructure metrics be computed live off an order-book stream? This project was built to find out: a FastAPI backend streaming Binance data through a stack of quantitative modules, with a React dashboard on top. Analysis only — never meant to trade real money.

What’s inside

Smart Money Concepts detection (order blocks, fair value gaps, break of structure, liquidity sweeps), AMD accumulation/manipulation/distribution heuristics, Order Flow Imbalance and volume-weighted micro-price over the live book, cointegration testing for pairs trading, and a backtest engine to run strategies against history. The dashboard is React + MUI with TradingView Lightweight Charts and socket.io streaming.

What I learned

Beyond the quant math: the value of an honest README. The repo now documents what it does not do — no auth, no real neural net behind the “transformer” module, order paths untested against live exchanges — because an exploration project earns more trust by stating its limits than by overselling. No longer developed; kept public as reference for the microstructure and stat-arb modules.