Weekly range volatility as a tactical regime filter

What range-volatility measures

Range-volatility is defined as the period high minus low, divided by the close and scaled by 100. Historical-volatility is that high-to-low move in a day or week, expressed as a percentage of that period's close, and is used to map the band a single issue, average, or index tends to occupy.

Each issue, average, and index is treated as having its own characteristic-volatility-band: the recurring span of daily or weekly range readings that belongs to that name rather than to the market as a whole. Range-volatility is also described as relatively constant over time and not always tied to volume.

The weekly-noise-threshold

Weekly readings below 10 percent are framed as lower-risk noise. The weekly-noise-threshold is a weekly range reading near or below 10 percent, treated as ordinary fluctuation that can persist. Weekly readings above 10 percent are framed as a possible pending price change. A volatility-expansion-regime is a stretch in which weekly range readings move above 10 percent and are treated as typically short-lived.

Weekly volatility above 10 percent is treated as typically short-lived, and weekly volatility below 10 percent as noise that can continue indefinitely. Daily volatility is described as about 45 percent of weekly volatility, which is the daily-to-weekly-ratio.

Two illustrated names

In a 1986 four-month sample, one quieter name showed daily readings from 1.06 percent to 4.74 percent with a 2.43 percent average, a 4.5 percent price swing from 26 to 22.125, and a 5.6 percent average weekly reading over the same high-low span.

In a 1988 sample, a more active name showed daily readings from 3.85 percent to 9.52 percent averaging 6.03 percent, a 15 percent price rise from 2.500 to 3.875, and weekly readings averaging 12.57 percent with a 6.9 percent to 24.14 percent week-to-week span.

Both illustrated names showed long stretches without large volatility swings, so a band can persist rather than reset every session.

Venues, a basket, and a screen

A one-week sample of roughly 4 percent to 5 percent of names on three U.S. venues, taken during an approximately 85-point market drop and about 195 million shares of daily volume, found more high-volatility names on the New York exchange and more extreme high-and-low readings among NASDAQ names.

Over a five-month weekly window in 1986, a thirty-name industrial basket averaged 6.3 percent weekly volatility versus 4.0 percent for the industrial average itself.

A screen of names priced under 10 with weekly volatility of 10 percent or more produced 250 candidates, and a smaller look at names priced between 5 and 14 showed usable readings at many volatility levels.

Regime clues on a single name

Market-regime-classification labels a stretch of tape as ordinary weekly noise, a typically brief expansion, a dormant awakening, a damped decline, or a late-move spike so one name is read in a broader context.

A rise after dormancy, a decline that stays volatile through a low, a decline that damps, and a late-move volatility increase were each framed as different regime clues for later price behavior, including the possibility of a top.