Lookback and direction-test parameters for a liquidity-plus-trend screen

Hidden defaults in a liquidity-plus-trend screen

A liquidity-plus-trend screen is not testable until two hidden defaults are written down. The first is the lookback used to form the moving-average of volume behind a liquidity-filter. The second is the direction-test that turns a moving-average of price into an up, down, or sideways call.

Editorial reading: those parameter choices, not the headline share-count floor or the named length of the price average, decide which names pass and whether a market is treated as trending or merely drifting.

The lookback behind the volume floor

A reader asked which lookback should be used to decide whether average daily volume exceeds one million shares, and listed two, five, twenty, forty, and sixty days as candidate windows. A moving-average of volume is a simple average of ordered volume observations over that stated lookback. A liquidity-filter then keeps only names whose recent average volume clears the stated share-count floor, so thinly traded issues are excluded before a signal is acted on.

A screen that required a twenty-day simple moving-average of daily volume above one million shares still returned names whose volume on the scan day was far below that floor.

A reply held that a lookback of twenty days or longer was not expected to change the outcome much, because the filter's purpose was to avoid thinly traded names. Twenty days was equated with four trading weeks, with the caveat that a low-volume day or even a low-volume week can still appear inside that window. The same reply argued that names clearing a twenty-day-or-longer volume floor should still have enough activity to be traded by institutions, even if a quiet day or week falls inside the average.

The direction-test on a fifty-day average

Another question asked what explicit test turns a fifty-day simple moving-average of a broad-market proxy into a direction call. It asked whether the latest reading must exceed the reading ten periods earlier, and whether a percentage threshold is required. That comparison is the direction-test, including any prior-period or percentage requirement.

Illustrated cases were used to show that a fifty-day average can rise while price remains range-bound, so an advancing average does not by itself distinguish a clear trend from a sideways market.

Editorial reading: trend-following converts a market-state reading, such as the path of a moving-average, into an entry, exit, or abstention decision for the system's holding period. That procedure stays untestable until the direction-test is written down beside the lookback.