How far the current figure is from its average, in standard deviation units. 1.5 or more = an event that happens less than 7% of the time.
What it measures
The Z-Score (Z-Score) answers: how many standard deviations (SD · Standard Deviation) is the current value from its average? Zero means exactly at the average, and +2 means two deviations above it. An event like that happens statistically about 2.5% of the time in a normal distribution. It's the common language of extremes across all assets and measures.
Why it's the quant's ruler
It makes the incomparable comparable: "a stock $7 above its average" means nothing until you know its volatility, while "+2.4 deviations" means the same thing for any asset. Anomaly filters are built on it: anything past a standard threshold deserves a look, whether volume, price, correlation or flow.
Its honest limits
Markets aren't a normal distribution. Their tails are fatter than theory assumes, so "three standard deviation" events happen more often than the table promises. The score ranks oddities and doesn't compute their probability precisely. Use it as a sorter, not a probability machine.
Hypothetical example
A symbol whose volume on alternative venues is used to 5 million shares a week with a one-million swing: a week of nine million logs +4 deviations. That exact anomaly is what the hidden volume radar catches before any explanation.
Common mistake
Applying it to a window whose character has changed: the average and volatility of a quiet period make the first stormy week look like a "superhuman anomaly". In truth the regime changed. The score belongs to its window.
Why it works for comparison
It turns any measure into "how many standard deviations from its average", so assets with different units can be compared on one ruler. Its limit is that it assumes a symmetric distribution, and markets have fat tails: a score of three happens more often than the model assumes.
Where to see it on the platform
The hidden volume radar ranks symbols by it, the anomaly filters in the flow tools are built on it, and it appears on the symbol card wherever a deviation is worth mentioning.
The platform's indicators show the value with its percentile position, so the number is read in context.
