How can information about Williams %R be verified?

Direct answer

Information about Williams %R can be verified by checking three things: (1) a stable definition of what it measures, (2) a reproducible calculation procedure, and (3) limitations that can change the displayed values even when the formula is correct. Because there is no guarantee that two platforms use the same inputs or window alignment, independent recalculation from clearly stated data requirements is the most reliable way to confirm claims.

Mechanism or definition

Williams %R (often written Williams %R) is a momentum oscillator. It is computed from a lookback window of recent price information: the highest high and the lowest low over that window, and a current price point (commonly the latest close). A typical presentation is that it is expressed on a fixed numeric scale (often described as being bounded between two extremes).

To verify the concept, separate stable mechanics from variable conditions:

Stable mechanics: the relationship between highest high, lowest low, and the current price point over a defined window.
Variable conditions: which exact price series is used (close vs another field), the lookback length, the alignment of the window to the current bar, and the data source.

Evidence or example (reproducible verification steps)

Use a controlled, offline check. The goal is to reproduce the indicator values from the same assumptions a provider claims.

Fix the assumptions

Choose a lookback window length (for example, N periods). State N explicitly.
Decide the current input price used in the formula (commonly the close of the current period). State it.
Use the same bar definition as the source you want to verify (e.g., same timeframe and timestamp conventions).

Collect the inputs For each calculation point, collect:

Highest high over the previous N periods (including the period you define for the window).
Lowest low over the previous N periods.
The current period’s chosen price (e.g., close).

Recalculate step by step Compute the numerator using the distance between the highest high and the current price point, then divide by the range between the highest high and the lowest low over the same window. Convert to the provider’s displayed scale only if the provider uses a scaling convention that is part of the definition.
Compare with the provider

Recalculate for multiple points, not just one.
If the values differ, treat it as a mismatch in assumptions: window alignment, choice of price field, timezone/bar construction, or handling of missing data.

Round-trip consistency test Verify that the indicator responds in the expected direction when you modify inputs in a controlled way:

If the current close moves toward the high of the lookback window (with other inputs unchanged), the computed oscillator should move accordingly under the formula. This does not confirm future prediction performance; it confirms internal arithmetic consistency.

Limitations and risks

Even if the formula is correct, verification can fail due to practical mismatches:

Division-by-zero or near-zero range: if highest high equals lowest low over the lookback window, the computation can be undefined or unstable. Different platforms may handle this differently (e.g., output a default value).
Window alignment differences: using “previous N bars excluding the current bar” versus “including the current bar” changes the highest-high/lowest-low values.
Data-source differences: different providers may build candles differently (timezone, session breaks, corporate actions for some instruments), leading to different highs and lows.
Missing candles or interpolated data: gaps can alter the effective window.

A material failure mode is comparing your recalculation to a platform output without confirming the platform’s exact lookback alignment and input price field.

Verification or next question

To verify any specific claim about Williams %R (including formula variants), confirm the claim’s exact assumptions: lookback length, which price field is used as “current,” how the window is aligned, and how special cases (like a zero range) are handled. If you want to go further, the next step is to cross-check the same historical dataset using two independent data sources and confirm whether differences are attributable to input construction rather than to the indicator’s underlying mechanics.