When quiet-day breadth fails a horizon test

A quiet-day running total

The Negative Volume Index is a market-breadth running total that updates only on sessions when volume is lower than the prior session, turning that quiet-day advance-decline residue into a directional signal.

The original rationale treats quieter sessions as the setting for professional positioning and high-volume sessions as the setting for public participation.

A later breadth comparison

A later construction first applies a weekday-volume-correction, a rescaling of daily volume so systematically quieter weekdays can be compared with the rest of the week, then applies 33 percent exponential averages to advances, declines, and total issues on declining-volume days and on all days.

Market breadth is used here as a comparison of advancing and declining issues, applied both on declining-volume days and across all days so the two streams can be contrasted. The published series compares those two smoothed market-breadth readings and then applies another 33 percent exponential average to the comparison.

Unusual readings across five horizons

An unusual-reading is a threshold that labels an indicator value as high or low when it sits beyond one standard deviation of the series. A 10-year directional check asked whether unusually high or low readings, marked with a one-standard-deviation band, anticipated the industrial average at one-week, five-week, 13-week, 26-week, and one-year horizons.

A chi-square test is a categorical association check that asks whether unusually high or low indicator bins line up with later up or down moves of a price benchmark over a stated horizon.

The conventional sign inverts

A chi-square test found a statistically notable association only at the six-month horizon, and that association ran opposite the conventional reading: high figures lined up with later declines and low figures with later advances.

Year-by-year tallies were unstable, including 1985 with no correct directional calls against eight incorrect ones and 1988 with eight correct calls and none incorrect.

A warning about conceptual appeal

The write-up treated the modern results as consistent with a 1969 judgment that the series had begun to produce false signals and framed the exercise as a warning against accepting the construction on conceptual appeal alone.