Evaluating a black-box pyramiding routine with adverse excursion

Audit the observed procedure

A mechanical trading system is a fully specified entry, exit, and abstention procedure that can be run as one testable routine from market inputs, state, and execution constraints. When the internal rules are withheld, the object is a black-box system, and evaluation rests on observed signals, fills, and excursion rather than on inspecting the logic.

An undisclosed mechanical futures routine can still be evaluated as one procedure of entry, reversal, and abstention once daily signals and execution constraints are observed.

Contrary stance and pyramiding

The observed stance often stays contrary to the prevailing move and adds contracts as the open loss grows, waiting for ordinary oscillation to restore a profit before reversing.

Pyramiding means adding contracts to an open stance in steps, including after the position has moved adversely, so total exposure is built rather than placed all at once.

Disk documentation stated the built position could reach 15 contracts, while the designer advised a four-contract cap that would require recomputing any reported record. The evaluation used the shipped defaults.

Daily bracket stops after a direction change

After a direction change, a buy stop and a sell stop bracket the first unit so a favorable trigger adds size and an unfavorable trigger reverses. Those bracket stops must be resized and relocated every day.

Internal logic is withheld. The observed horizon is consistent with a pricing-pattern routine that typically uses two to five days of open-high-low-close data to enter and holds two to fifteen days, often with added units.

Quote integrity belongs in the procedure

A one-tick change in the opening or closing print can switch a signal from an at-open order to a paired stop recommendation, so quote integrity and data-entry checks are part of running the procedure.

Compare winner and loser excursion

Maximum adverse excursion is the largest opposing price movement observed while a trade is open. It is used to compare winning and losing trades and to bound loss behavior before and during a position.

Maximum adverse excursion for winning trades and losing trades was generally small, and the two curves did not differ in a meaningful way, which is consistent with relying on later oscillation rather than on conventional winner-versus-loser entry selection.

Adding contracts after adverse movement increases the count of trades that later close on a bounce, so a high share of favorable outcomes cannot be treated as independent proof of better entries.

The inspected window is too short

Security features blocked applying the program to older data, so independent historical verification was unavailable and only a nine-month record could be inspected.

Combining three contract months over 136 days produced 65 trades. That short sample does not show how the add-to-losers logic behaves in a strong one-way trend.