For each period in a historical lookback, the net change is measured as a percentage return from the open, calculated as (close − open) / open. Collecting these returns across hundreds or thousands of periods produces a robust distribution of outcomes. The sample is then used to calculate a ±1 standard deviations.
The standard deviation of that distribution quantifies the typical magnitude of movement. Multiplying the standard deviation by the current open price converts it into concrete price levels, producing bands above and below the open at any fraction of ±1 standard deviations.
Assuming returns are roughly normally distributed, each band corresponds to a statistical probability by converting zscore to percentile. The bands map out where price has usually stayed, and where reaching becomes progressively rarer.
The upper figures are the two-sided probabilities of closing anywhere inside the ± band. The lower figures are the one-sided probability of closing on one side of a single SDEV value — e.g. below the +1 SD level (or, symmetrically, above the −1 SD level).
The information is treated as a probabilistic map of expected range rather than a signal system. Because most periods close within ±1 standard deviation, extensions to the outer bands are statistically stretched and often watched for mean reversion — price fading back toward the open. A decisive push beyond a band instead signals an unusually strong period, sometimes used as breakout confirmation where the market may create an outlier event, dependent on that bands percentile value.