Quant

An async feed handler with auto-reconnect, an in-memory order book reconstructor, a lakehouse that rolls ticks into OHLCV candles in Parquet and DuckDB, NumPy-vectorized indicators written without a TA library, and a pluggable strategy engine that emits strict entry and exit signals.

Signals are replayed second by second in an event-driven backtester, with latency and slippage injected by a market simulator, checked by a risk gate (position sizing, stop-loss, max drawdown), routed to a live exchange or paper trading, measured (Sharpe, Sortino, profit factor) and logged to a millisecond audit trail — all monitored from a real-time web terminal.