Define the concept before the math
“RSI Range” is a way to express where the Relative Strength Index (RSI) currently sits relative to its own recent extremes. Instead of using the RSI as a single number only, RSI Range maps the current RSI value onto the range formed by the RSI’s recent minimum and maximum.
To calculate RSI Range you need to:
- compute RSI from price data using a chosen lookback length (the RSI period), then
- take a second lookback window (the RSI Range window) and find the minimum and maximum RSI values over that window,
- convert the current RSI into a normalized position within that [min, max] interval.
This keeps the mechanics consistent; only your chosen parameters control the behavior.
Mechanics: inputs, parameters, and the formula
Step 1: Compute RSI from price
RSI is typically defined using a period length (n) (often 14). Using closing prices, RSI is based on average gains and average losses over that period.
A common formulation expresses it as:
- Compute gains (G) and losses (L) for each step (where (G) is positive when price rises; (L) is positive when price falls).
- Compute average gain (\overline{G}) and average loss (\overline{L}) over the RSI period (n).
- Then: [ \text{RSI} = 100 \times \frac{\overline{G}}{\overline{G} + \overline{L}} ]
If your implementation uses a different RSI averaging method (for example, Wilder-style smoothing versus a simple average), the computed RSI series can differ, and therefore the resulting RSI Range can differ too.
Step 2: Define the RSI Range window
Choose a second parameter: (m), the RSI Range window length. At each time (t), you look back (m) periods and compute:
- (\text{RSI}{\min} = \min(\text{RSI}{t-m+1}, \dots, \text{RSI}_t))
- (\text{RSI}{\max} = \max(\text{RSI}{t-m+1}, \dots, \text{RSI}_t))
Step 3: Convert current RSI into a normalized “range position”
A typical RSI Range calculation is a min-max normalization of the current RSI: [ \text{RSI Range}t = 100 \times \frac{\text{RSI}t - \text{RSI}{\min}}{\text{RSI}{\max} - \text{RSI}_{\min}} ]
Interpretation of this value:
- If (\text{RSI}_t) equals the minimum RSI in the lookback window, RSI Range is 0.
- If (\text{RSI}_t) equals the maximum RSI in the window, RSI Range is 100.
- Values between min and max place RSI proportionally within that interval.
Required data and assumptions
To compute the numbers, you must have:
- consistent price series (commonly closes),
- a defined time step (for example, daily bars or hourly bars),
- the two parameters: (n) for RSI period and (m) for RSI Range window,
- and a consistent RSI method (smoothing/averaging rules).
If any of these differ between two implementations, the RSI Range series can differ even when the underlying market is the same.
Evidence or example: a self-check you can verify
Below is a simple, verification-oriented example using hypothetical RSI values.
Example using RSI values (skip RSI computation)
Assume that for a given time (t), you have already computed RSI values over the last (m=5) periods:
- (\text{RSI} = [42,\ 46,\ 40,\ 48,\ 45])
Then:
- (\text{RSI}_{\min} = 40)
- (\text{RSI}_{\max} = 48)
- (\text{RSI}_t = 45)
Apply the formula: [ \text{RSI Range}_t = 100 \times \frac{45 - 40}{48 - 40} = 100 \times \frac{5}{8} = 62.5 ]
This tells you how to check your implementation without guessing: when RSI is halfway between the min and max, the RSI Range will land halfway between 0 and 100.
Example using prices (what you must still ensure)
If you start from prices instead, your self-check is to confirm that:
- your RSI output uses the same RSI averaging method and period (n),
- the RSI Range window (m) is applied to the RSI series (not prices), and
- your min and max are computed over exactly the same set of (m) RSI observations.
A frequent failure mode is accidentally using (m) periods of price extrema (or using an off-by-one window), which changes (\text{RSI}{\min}) and (\text{RSI}{\max}), and therefore changes the normalized value.
Limitations and risks: material failure modes to expect
1) Division by zero (or near-zero range)
If (\text{RSI}{\max} = \text{RSI}{\min}), then the denominator ((\text{RSI}{\max} - \text{RSI}{\min})) becomes zero, and the formula is undefined.
This can occur when RSI is flat over the entire lookback window, or when the computed RSI values are identical due to rounding, data issues, or an implementation that produces repeated values.
Material implication: your software must define what happens in this case (for example, returning a default value or marking it as undefined). Different choices lead to different plotted results.
2) Sensitivity to parameter choices
RSI Range behavior depends strongly on (n) (RSI period) and (m) (range window). Larger (m) uses older RSI extremes, which can make the normalization more stable but less responsive. Smaller (m) makes it more sensitive.
Therefore, two parties can compute different RSI Range values from the same price series if their parameters or RSI method differ.
3) Implementation differences in RSI averaging
RSI is not one unique computation in practice. Common RSI implementations can differ in whether they use:
- Wilder-style smoothing,
- simple moving averages,
- or other averaging conventions.
Even if both implementations use the same (n), the RSI time series can differ slightly, and the derived RSI Range can shift.
4) Data quality, time step, and corporate actions
RSI and the RSI Range are only as reliable as the underlying input series. Practical issues include:
- missing bars,
- inconsistent time zones or trading hours,
- and data source differences.
These issues can create artificial RSI movement, which then changes (\text{RSI}{\min}) and (\text{RSI}{\max}).
Verification and next question to ask
How to independently verify your RSI Range
To verify a computed RSI Range series:
- Recompute RSI exactly the same way (same period (n), same averaging method).
- For a chosen time (t), list the last (m) RSI values and compute (\text{RSI}{\min}) and (\text{RSI}{\max}).
- Plug (\text{RSI}t), (\text{RSI}{\min}), and (\text{RSI}_{\max}) into the normalization formula.
- Confirm the result matches the indicator output at time (t).