Upgrade High-Speed Metals Trading Bot

I have an automated trading bot that already runs in production, but I now need it to perform flawlessly on the metals market—specifically gold and silver—and to do so at lightning speed. The strategy relies on fast short-term trades, so every millisecond trimmed from order routing and every extra layer of safety added to position sizing makes a measurable difference to P&L.

Here is what I expect from you:

• Refactor or rewrite latency-sensitive components to improve execution speed on metals exchanges (low-level socket/IP tweaks, multithreading, asynchronous order handling—whatever it takes).
• Build a configurable risk-management module: max loss per session, dynamic position limits tied to volatility, and automatic circuit-breaker logic when preset thresholds are breached.
• Run thorough historical backtests using at least five years of tick or 1-minute gold and silver data, validating edge persistence over multiple market regimes.
• Put the enhanced bot through live paper trading for a defined burn-in period, delivering detailed performance and slippage reports.

I code-review everything, so please use clear, well-commented structures. The existing stack is mainly Python with some Cython speed-ups; feel free to propose smarter alternatives (C++, Rust, GPU acceleration, etc.) if they meaningfully reduce latency without compromising stability. Final deliverables are:

1. Optimised and fully tested source code
2. Risk-management module integrated and documented
3. Backtest notebooks or scripts with reproducible results
4. Paper-trading logs and a concise performance summary

Acceptance criteria: average order round-trip must beat the current baseline by at least 20 %, and the risk controls should pass my supplied stress-test scenarios unaltered.

If you thrive on squeezing out microseconds and bullet-proofing trading systems, let’s talk.