admin/Trademind_OpenCode
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

1 Release 1 Tag

RSS Feed
  • v1.0.0 7959dd71ff
    Compare

    Version 1.0.0 Stable

    admin released this 2026-08-22 11:54:09 +02:00 | 1 commits to main since this release

    Highlights:

    • Three operating modes — single CLI entry point:
      • trademind paper — simulation with real market prices (public ccxt API), orders only simulated
      • trademind live — real orders via ccxt (Binance, Kraken, Coinbase, KuCoin, OKX, Bybit, …)
      • trademind train — evolutionary optimization of strategy parameters and signal weights
    • Core strategy: MACD cross + RSI filter + ATR-based stop-loss, long/short configurable
    • Strategy training: walk-forward mode with elitism, crossover & mutation; fitness = 0.6×Return + 0.3×Sharpe/10 − 0.1×MaxDrawdown; trained weights persisted to state/weights.json and auto-loaded by paper/live
    • Data sources: --data auto (real market → fallback to deterministic mocks), --data live (strict), --data mock (offline, reproducible), exchange selectable per run
    • Deployment: single-file Podmanfile + optional podman-compose.yml; API keys via env vars, config template included (no secrets committed, state/ gitignored)
    • CLI: trademind show to inspect the effective configuration with masked keys
    • Quality: pytest suite (7 tests) passing; dependency floor pinned in requirements.txt (ccxt, pandas, numpy, PyYAML, click)

    Getting started:

    podman build -t trademind:latest -f Podmanfile .
podman run --rm trademind:latest paper --config /app/config.yaml

    Notes & disclaimer:

    • Default config runs in sandbox/paper mode (sandbox: true, dry_run: true).
    • Live trading is at your own risk — no guarantee of returns.
    Downloads