Direct answer
A worked example of mean reversion range is a numerical scenario that shows how you (1) choose a reference average, (2) estimate typical variability around that average from historical observations, and (3) define a “range” as fixed distances above and below the average. You then compare a later observed price to that range to understand what “within range” and “outside range” mean—without assuming it predicts future outcomes.
Mechanism or definition
Mean reversion is the idea that, over some lookback window, a variable may tend to move back toward a central tendency (an average). A mean reversion range turns that qualitative idea into a rule that produces levels.
A common worked-example setup uses four assumptions that you must state explicitly:
- Reference average: you compute the average of a chosen quantity (often a price level or a derived series) over a lookback period.
- Variability measure: you compute how much the series typically deviates from that average (for example, using standard deviation).
- Range width: you pick a multiplier k that scales variability into distances.
- Comparison moment: you use a separate “current” value to test whether it lies inside or outside the defined range.
Stable mechanics are the arithmetic (average, variability, and adding/subtracting a scaled variability). Variable conditions are everything else: the chosen lookback window, the definition of the series, and whether market behavior remains similar to the historical period. Since no real-time data or provider-specific settings are assumed here, the example focuses only on method and interpretation.
Evidence or example
Worked numerical scenario (all assumptions stated)
Assumptions (fixed for this example):
- You use 10 historical observations of a single series value, labeled (x_1) to (x_{10}).
- The series is already expressed in the same units consistently over time.
- You compute the arithmetic mean (\mu) across the 10 values.
- You compute the sample standard deviation (s) from those 10 values.
- You set the mean reversion range as ([\mu - k,s,; \mu + k,s]) with (k = 1).
- You then compare one “current” observation (x_{current}) to those two levels.
Historical values (10 observations): [ [1.1000,;1.1010,;1.0990,;1.1020,;1.1005,;1.0985,;1.1015,;1.1002,;1.0998,;1.1012] ]
Step 1: Compute the average (\mu). The sum of the 10 values is (11.0027). Therefore: [ \mu = 11.0027 / 10 = 1.10027 ]
Step 2: Compute the sample standard deviation (s). Using the sample standard deviation formula (divide by (n-1)), the result for this set is approximately: [ s \approx 0.00119 ]
Step 3: Set the range with (k=1). Lower level: [ \mu - s = 1.10027 - 0.00119 = 1.09908 ] Upper level: [ \mu + s = 1.10027 + 0.00119 = 1.10146 ]
So your mean reversion range is approximately: [ [1.09908,; 1.10146] ]
Step 4: Compare a current value. Let (x_{current} = 1.1022). Since (1.1022 > 1.10146), this current value is outside the upper boundary.
How to interpret the result (without predicting)
From this computation alone, you can only say:
- The range was defined from historical variability, not from guarantees.
- The current observation lies outside the historical band as defined by your chosen (\mu), (s), and (k).
- Whether the series later moves back toward (\mu) is unknown and depends on changing market conditions.
Limitations and risks
- Definition risk (lookback and series choice): Changing the lookback window, data granularity, or the exact series definition changes (\mu) and (s), which changes the range.
- Regime shift / non-stationarity: Mean-reversion behavior can weaken if the process generating values changes. In that case, historical averages may no longer represent the current regime.
- Outlier sensitivity: If your variability estimate is influenced heavily by a few extreme points, the range can become too wide or too narrow.
- Costs and execution (in real trading contexts): Even if the computed range is mathematically correct, real outcomes can diverge due to bid/ask spreads, slippage, commissions, and latency. (Those factors are not included in this worked example because no provider or market conditions are assumed.)
- Verification mismatch: A successful “backtest” over one period does not establish future performance; historical relationships do not guarantee future results.