Constructing a mechanical crossover on a synthetic price series

A bench that invents the series

An object-oriented programming language was used as the construction environment for market models, stock charts, and rule-driven evaluation. Object-oriented modeling here means representing prices, charts, and rule evaluators as reusable program objects rather than as one-off spreadsheet cells.

A moving-average crossover turns a shorter average crossing a longer average into an explicit long, short, or flat signal. The archive placed that rule inside rule-driven evaluation: coded signal rules applied to a series so results can be plotted, exported, or optimized as one procedure.

What the environment supplied

That language was judged relatively unfamiliar to most computer-based market technicians at the time of writing. The environment was presented as cheaper than typical analysis programs, with a retail offering below 100 dollars, and as comparatively simple to learn.

Built-in graphics produced the stock charts, which were captured from a 286-class computer with a CGA display. EGA or VGA was described as enabling full-color charts.

From plot to file to search

Aside from some analysis performed in a spreadsheet, the modeling and plotting work was done in no more than 100 lines of code. Sending generated series to a file instead of, or in addition to, plotting them was described as requiring only four or five lines of code. A small extra method could write those files in a then-common technical-analysis package format.

Synthetic price series were generated in that environment. Those series are computer-generated pseudo-market data used to build and test rules independently of any listed security. They were then examined with subscriber analysis software, including a search for a moving-average crossover setting that treated profit on the invented series as the historical objective. That step is system optimization: a search over rule parameters on a fixed series to locate settings that maximize a chosen historical objective.

A mechanical trading system is a single procedure that binds entry, exit, and abstention to stated inputs so the whole path can be rerun without discretion. The same generated series was plotted, written to a file, and examined in subscriber analysis software.