AUTONOMOUS TRADING SYSTEM

Most automated traders are confident long before they are correct. They go live on a backtest that looked good and a risk policy that lives in a readme file, and the first bad day takes a real piece out of a real account before anything stops it.

On the ninth of June, the hand-rolled first version of this system lost 7.4 percent intraday with no circuit breaker. The four risk rules it claimed to follow were written in documentation and never enforced anywhere in the execution path.

That day is the reason the whole thing was torn down and rebuilt around one rule: a strategy does not get to be fast until it has proven it can be safe.

Shadow mode is the load-bearing idea. A new strategy does not start on real capital and it is not judged on a backtest. It trades against a live paper account, scored on real prices as the market moves, and its results are tracked separately where they cannot deploy a dollar.

Graduating does not mean money either. It means the system has earned the right to ask. Six conditions have to clear first.

The lesson from the bad day was that a risk rule you only wrote down is not a risk rule. So four hard limits are enforced as code that every order has to clear before it reaches the broker.

No order may risk more than 2 percent of the account if its stop is hit. No single name may grow past 25 percent of the account. The account may never spend its way below a 10 percent cash floor. And if the account is down 3 percent on the day, a breaker trips and blocks all new buying while still always allowing the system to sell and reduce risk.

The four checks are one self-tested function with no side effects, built that way so it can never be quietly bypassed. There is also a manual pause a person can flip to freeze all buying. The rules that protect the money are not a promise in a file, they are a gate in the path.

Two strategies feed the router. One follows price momentum. The other follows the stock transactions of members of Congress, which are required by law to be disclosed publicly under the STOCK Act.

That data is not a private feed and it is not bought from anybody. It is read directly from the government disclosure sites for zero dollars, the same source subscription apps quietly resell.

The edge is not speed. By law those disclosures can lag the actual trade by up to 45 days, so nobody is front-running anything. The edge is that almost no retail investor acts on the signal, and that the system ranks each filer by their own historical record, so a disclosed buy from a consistently strong filer carries more conviction.

It would have been faster to keep hand-rolling. Instead, after the bad day, I deleted roughly 76 kilobytes of my own trading code and adopted freqtrade, the most widely deployed open-source retail trading framework there is, with a risk module thousands of operators have beaten on instead of one.

I kept the two things that were genuinely mine: the congressional disclosure signal feed, and the reliability rails that alert a person the moment anything fails.

That is the same judgment I bring to client work. Do not rebuild the engine, adopt the proven one, and spend the effort on the part that is actually your edge.

This is a personal system, paper only, held in shadow by design. It will not move real capital until every condition has cleared and I have personally replied go live. When a strategy graduates, the system fires a notification and waits.

It is on this site because it is the clearest example of how I build anything unattended. AI is the engine that lets one marketer run systems at this size, and the discipline is what keeps the engine from being dangerous.

The same three habits show up in every automation I point at a business. Enforce the guardrails in code rather than in a document, prove the thing on a live surface with nothing at stake, and keep a person on the switch that matters.

I crawl your whole site, score it 0 to 100, and send a ranked list of what is costing you customers. About a day. No call needed.

If the fix is not something I do, I will say so.