A construction test for a modified volume-price trend filter

What the construction targets

The construction targets interday staged buying or selling of a large stake. It uses only price and volume. It compares prints across successive days rather than the same day's open and close.

The full-volume filter

A full-volume cumulative filter starts at the first bar's volume and then adds or subtracts each later bar's entire volume according to the sign of the close-to-close change, even when that change is tiny.

Because every bar contributes its full volume, that filter is typically choppier than price. The extra movement can hide a genuine flow imbalance or suggest one that is not there.

The volume-price increment

A volume-price increment applies only a share of the bar's volume in proportion to the fractional price change. This increment removes much of the day-to-day jumpiness of the full-volume series.

Replacing the close

A moving average here is a same-bar average of open, high, low, and close. It replaces a single close in the increment and is compared with the prior bar's average. Replacing the close in this way pulls the volume-price path closer to price when no staged flow is present.

Matching the filter to log price

The modified overlay plots price on a logarithmic scale and the trend filter on a linear scale. A level shift and a scale factor are then applied so leftover slope can be inspected after alignment. The trend filter is a cumulative, volume-weighted price-change series aligned to price so residual slope can be read as staged buying or selling.

When every volume observation is replaced by the constant 1, a linear-on-linear overlay still leaves the paths far apart. Log price against a linear filter produces a near match.

Reading leftover slope

During residual slope disagreement, the construction treats the direction of the modified filter as the earlier reading for later price direction.

Region breaks

A sharp gap on unusually large volume is treated as a region break. New level and scale factors are fitted only between breaks. Flattening those jumps would blunt later slope disagreement.

After a market-wide crash

After a market-wide crash, leftover slope at the individual-stock level can become scarce. A common market force may dominate local series, or staged campaigns themselves may become rarer.