Ask someone to show you their trading strategy and they’ll show you a file. A class, a few signals, an entry and an exit. It’s the part that’s fun to work on, so it’s the part that gets the attention.
But the strategy file is maybe a tenth of a trading system. The other nine-tenths is everything that has to be true for that file to run, place real orders, and not quietly lose money while you’re not looking.
The part nobody screenshots
Here’s what sits between a strategy and a live trading system:
- An engine that feeds the strategy market events in the right order, without leaking the future into the past.
- A fill model that doesn’t pretend every order gets the last printed price.
- Market calendars, so the strategy isn’t trading a closed market or fumbling an early close.
- A broker integration, and a clear answer to what’s true when your software and the broker disagree.
- Orders, fills, positions, and a ledger you can actually inspect.
- Logs and structured errors, so when a run breaks at 2pm you can find out why.
- Somewhere to run all of it that stays up.
None of that is the strategy. All of it is the trading system.
Why this is where research dies
A backtest that works in a notebook has quietly assumed most of that infrastructure away. Perfect fills. No calendar. No broker state. No monitoring, because nothing is live to monitor.
When you try to take that same notebook to live, every assumption you skipped comes due at once, usually as a strategy that “worked in research” and does something baffling in production. The strategy didn’t break. The system around it was never built.
The honest framing
This isn’t an argument that strategy code doesn’t matter. It’s the thing only you can write, and it’s where your edge lives. The point is narrower: the strategy is the part you should be spending your time on, and the system around it mostly isn’t. The fill model, the calendars, the diagnostics, the compute, that’s undifferentiated work that looks the same for almost everyone who does it well.
So the real question isn’t “is my strategy good.” It’s “do I have a trading system, or do I have a strategy file and a pile of work I haven’t done yet.” Be honest about which one you’ve got, and then decide whether building the other nine-tenths is the best use of your time.