Constructing the slow volume strength index from signed volume

A volume analogue of the relative strength index

The slow volume strength index is a 0-100 oscillator that assigns volume to the up or down side according to whether the close is above or below a short exponential moving average, then applies Wilder-style averaging and the relative strength index transform.

The construction is presented as a volume analogue of Wilder's relative strength index in which the usual close-to-close change is replaced by the close-versus-moving-average comparison that signs the volume. The result is a volume-weighted RSI: an RSI-style ratio that uses classified volume, rather than close-to-close price change, as the signed input.

How each bar's volume is signed

The oscillator classifies each bar's volume as positive, negative, or unused by comparing the close with a short exponential moving average of the close, then averages those two volume series.

That exponential moving average is a recursively updated average of closing prices used here only to decide whether each bar's volume is treated as positive or negative.

When the close equals the exponential average, both the positive-volume and negative-volume inputs for that bar are set to zero.

Wilder-style smoothing and the bounded transform

After the first bar, both positive and negative volume averages are updated with Wilder-style recursion that multiplies the prior average by 13, adds the new classified volume, and divides by 14. That update matches the averaging convention of the relative strength index.

The intermediate ratio is defined as average positive volume divided by average negative volume when the negative average is not zero, and is otherwise set to 100. The oscillator is then 100 minus 100 divided by one plus that ratio, provided the ratio itself is not zero.

That last step is the relative strength index transform. It maps the positive-to-negative average ratio onto a bounded 0-100 scale.

Default lengths and reference levels

Default construction parameters shown in the implementations are an exponential-average length of 6, a smoothing length of 14, an upper reference of 80, a lower reference of 20, and a midpoint of 50.

Interpretation rules supplied with the implementations treat readings above 80 as overbought, readings below 20 as oversold, and also mention midpoint crosses and divergences as additional chart hypotheses. Those overbought and oversold bands, together with a centerline crossover through 50, are the supplied reference readings.

Sample rule sets and the shared formula

One sample rule set uses a close above a medium-term simple moving average as the trend filter, an oscillator print below 50 as the dip condition, and a next-bar market entry as the execution step.

A second sample rule set buys the next bar at market when the oscillator crosses above 50 and sells short the next bar at market when it crosses below 50.

The same volume-split, Wilder-smoothed construction is restated across multiple platform implementations, including TradeStation, eSignal, and Updata, so the teaching object is the shared formula rather than any one vendor listing.