Monthly eight-bar average entry and exit rules

What the illustrated procedure did

The illustrated procedure bought when a monthly close was above an eight-bar average and sold short when that close was below the same average, provided the account was not already on that side.

That comparison used a moving average as a lookback average of ordered monthly prices, the explicit quantitative baseline set against the latest close. Rule-based entry went long only when the close was above the chosen average and the account was not already long. It could go short only when the close was below the average and the account was not already short.

How the system was coded

The coded strategy accepted the price field to average, an eight-bar default lookback, and a switch among simple, exponential, and weighted averages as explicit inputs.

Buy and sell-short orders were placed at the next bar's market only after the position check. The implementation raised an error unless the chart used monthly bars. Those constraints made a mechanical trading system: a fully specified procedure that converts a close-versus-average comparison into long, short, reverse, or exit actions, including next-bar execution and a required monthly bar interval.

Other recodings of the same comparison

Another implementation treated the eight-period monthly average as a crossing signal line, hid older markers with a trailing-bar display limit, and recorded clustered reversals after a late-May 2007 sell, then a later stretch in which price oscillated around the average.

A further recoding compared an eight-month average of monthly price to that price and issued buy or sell orders only when a last-day-of-month condition coincided with a cross. It offered additional average types beyond simple, exponential, and weighted.

A daily-bar implementation read the monthly close from a paired monthly series, defaulted to an eight-period average with longs on and shorts optional, and entered only when the daily close matched that month-end close.

Calendar day, lookback, and a long-only hold

One test kept the comparison rules fixed but rebuilt the monthly average on each calendar day from 1 through 28 as well as at standard month-end, and separately varied the lookback from 3 to 21.

A companion write-up described staying long while the monthly close remained above the eight-month average and exiting when it broke below. It plotted equity and drawdowns for 2003-11 and 1990-2002, and it also plotted how a risk-adjusted ratio changed with average length from 1990 through 2011.