Five coordinated bots trading index ETFs and large-caps, fed by one capital allocator with real-time P&L tracking.
View source — prometheus-algo-traderFive separate bots run side by side. A central allocator decides how much capital each one gets and rebalances as conditions shift, so the strongest setups get funded and the rest stand down.
A market-regime check sits in front of everything — when the environment looks hostile, the engine throttles down rather than forcing trades.
Runtime: Python services coordinated by a capital-allocator module, run as systemd jobs; a small FastAPI service exposes the dashboard data.
Data & execution: equities data in, orders to an Alpaca paper account out, with a real-time P&L tracker persisted to SQLite.
Resilience: watchdog auto-restart, crash recovery, and a daily P&L report.
Telemetry: publishes a stats.json snapshot the front end polls; Telegram for alerts.
The shape of the code — enough to show how it is engineered, with the actual strategy logic redacted. The edge stays private.
# equity_engine.py — illustrative structure (strategy logic redacted) def allocate(bots, equity, regime): if regime.hostile: return stand_down(bots) # throttle # --- proprietary allocation weights: REDACTED --- weights = allocator.solve(bots, equity) for bot, w in weights.items(): bot.set_budget(w * equity) bot.step() # place/manage trades pnl.update(); telemetry.write("data/stats.json")