What runs where

The harness is split down the middle, and the split is on purpose. Research and execution are different kinds of work, so they run on different sides. Python is the lab. Rust is the engine. The terminal is where a person sits and watches both. Putting each job on the side where it belongs is most of what makes the whole thing hold together.

Python is the lab because the lab is allowed to be slow. This is where ideas get built, modeled, backtested, and put through walk-forward validation on data the model has never seen. It costs honestly for slippage and fees, because an edge that only survives free trading isn't an edge. Most ideas die here, and that is the point — the lab's one job is to decide whether an advantage is real before a single dollar is at risk. It can afford to be patient and thoughtful precisely because it never touches the market directly.

Rust is the engine because once the lab is done thinking, the strategy has to act. Market data comes in, risk gets checked, orders go out — the same way every time. We wrote it in Rust for a reason: it's fast, it's deterministic, and it can't be talked out of its own risk limits. Kill-switches are kept close. It doesn't improvise. It executes what the lab already proved, and nothing else.

The terminal is what the human sees. Instead of ten dashboards to babysit, there's one cockpit: positions, risk, and the handful of decisions that actually need a person. The point isn't to automate the trader out of the loop; it's to clear away the noise so the calls that matter are the only ones in front of them.

Fast where it must be, patient where it thinks, a human wired in. That's the whole shape of it — and it's how one person, or a small desk, can carry a book that used to take a room full of people.