Treat mechanical system design as a revalidation lifecycle

Archive workflow and editorial reading

The archive describes a historical workflow for building and supervising mechanical trading systems. TradersWeek editorial: read that workflow as a lifecycle of a precommitted exit, live-fill verification, and periodic revalidation, not as a search for one rule set that is meant to stay live without being turned off.

A mechanical trading system is a complete, pre-specified procedure that converts market inputs into entry, exit, or abstention decisions and can be tested as one unit. Walk-forward analysis treats the same rule set as a living procedure rather than a one-time backtest. Monte Carlo simulation turns an ordered series of historical outcomes into a distribution of possible equity paths so risk and robustness can be compared with later live or out-of-sample results.

The quitting point is part of the procedure

The archive workflow treats a mechanical trading system as one testable procedure only when its entry, exit, and abstention rules include a precommitted quitting point before any live capital is allocated.

A quitting point is a drawdown, streak, or duration limit chosen before live use. It turns a system off without waiting for a narrative explanation of failure.

Stopped systems stay on a monthly incubator watchlist

After a quitting point is hit, the same system can stay on a monthly incubator watchlist. It is considered for reactivation only if it later returns toward its long-term average.

An incubator watchlist is a large set of candidate systems observed on a fixed schedule so only a subset is live while others remain eligible for later reactivation. Monitoring more than 100 candidate systems can be reduced to a once-a-month comparison of each strategy performance report and equity curve rather than continuous live trading of the whole set.

Live fills decide whether the simulation is usable

A backtest is treated as live-achievable only when the same class of fills appears in real trading. A gap such as five simulated fills versus two live fills is a stop-the-system signal.

Lower-timeframe and nonstandard bar constructions, including tick, renko, and point-and-figure charts, are called out as cases where simulated fills often fail to match live execution.

Micro-sized contracts can be used as a live-debug layer. Signals may be taken from a larger, more liquid contract stream while execution and inexpensive live troubleshooting occur on the smaller contract.

Canned indicators remain unfinished inputs

A historical indicator study found standard tools were directionally correct only about half the time. Canned indicators are reframed as inputs that need nonstandard thresholds or a confirming price pattern.

Walk-forward and Monte Carlo come after the idea session

Walk-forward analysis is a rolling in-sample optimization and out-of-sample test sequence. Monte Carlo simulation is the later resampling step that compares a distribution of possible equity paths with later live or out-of-sample results.

Those steps belong after idea generation. A day-long workshop may spend only 15 to 20 minutes on ideas and then require repeated monthly production of systems that survive later validation.