Lock the TRIX construction before reading a zero-line cross

A programmable series for interval tests

A practitioner asked for a programmable TRIX strategy so that daytrading rules could be backtested across multiple sampling intervals.

In the archive workflow, TRIX is a constructed momentum series formed by stacking exponential moving averages of a chosen price input, then reading the result against a zero reference.

Default input, lookback length, and the zero line

One platform formula constructs TRIX as a triple exponential average of a selected price series, defaulting the input to the close and the lookback length to nine observations.

That same formula plots a zero reference line and treats a cross above zero as a turn positive and a cross below zero as a turn negative. That sign change is the zero-line cross used as a directional forecast event.

When reconstructions of the same market disagree

A spreadsheet reconstruction using a nine-observation average on a Nasdaq-100 tracking series broke the zero line in late August, while the published nine-period chart was described only as drifting toward zero.

From early October through late December the published series stayed above zero, while two independent reconstructions crossed zero several times. The letter writer judged that enough to function as opposing indicators.

A nested rebuild that reproduced the published chart

Rebuilding the series as a nine-observation exponential average of closes, then taking successive exponential averages of the prior average, reproduced the published chart on two platforms. That rebuild is the nested exponential average: an exponential moving average applied to the output of the previous exponential average until three smoothings are stacked.

Matching a third vendor's TRIX plot required shortening the period length to four observations rather than nine. Lookback length is the number of sampling intervals inside each exponential average, and changing it can reverse the sign of the same price history.

Construction disagreement is first-order

The letter listing the mismatch treated data source, calculation method, and user error as competing explanations, making construction details first-order rather than cosmetic.

When TRIX plots of the same market fail to match because the data, nesting sequence, or period length differ, that case is a construction disagreement.