The Problem

The Split-Brain
Dilemma

Research code ≠ Production

Rewriting Python notebooks into C++/Rust creates logic drift and buries hidden assumptions.

Venue quirks leak logic

Handling exchange-specific edge cases creates "adapter spaghetti" inside core strategy code.

Unverified Backtests

Simulations aren't auditable. Live incidents are nearly impossible to replay deterministically.

Research.py

Backtest: +40%

Logic Drift

Live Bot (Rust/C++)

Panic: API Change

Adapter Spaghetti

02