Rolling Fn

TransformUtility

Rolling Fn applies one of 30+ statistical functions over a sliding window of configurable length. At each bar, the function sees the most recent period values. Bars with fewer than period prior values produce null.

Algorithm

// For each bar i in [windowStart, len):

window=input[i − period + 1 .. i]

result[i]=fn(window) if i − period + 1 ≥ 0, else null

The window always has exactly period elements (or fewer near the start, which yields null). This makes it equivalent to a standard rolling indicator like a moving average.

Parameters

Name	Type	Default	Description
func	WinFuncType	'max'	Statistical function applied to each rolling window. See the full list below.
period	number	20	Rolling window length in bars. Bars before index period − 1 produce null.

Supported Functions

Descriptive Statistics

maxHighest value in the window.

minLowest value in the window.

averageArithmetic mean.

medianMiddle value when sorted.

sumTotal of all non-null values.

countCount of non-null values.

rangemax − min.

Dispersion

stdSample standard deviation.

varianceSample variance.

madMean absolute deviation.

cvCoefficient of variation (std/mean).

rmsRoot mean square.

iqrInterquartile range (q3 − q1).

q125th percentile.

q375th percentile.

Trend & Regression

slopeLinear regression slope over the window.

interceptLinear regression intercept.

r_squaredCoefficient of determination.

momentumLast value − first value of the window.

accelerationChange in slope between first and second halves.

Risk & Return

gainSum of positive changes in the window.

lossSum of negative changes in the window.

volatilityAnnualised volatility of returns.

sharpeMean return / std of returns.

sortinoMean return / downside deviation.

max_drawdownPeak-to-trough maximum loss.

Shape & Position

skewnessAsymmetry of distribution.

kurtosisTail weight of distribution.

argmaxIndex of maximum value within the window.

argminIndex of minimum value within the window.

Information

entropyShannon entropy (bin-based).

autocorrLag-1 autocorrelation.

rsiRelative Strength Index of the window.

Inputs & Outputs

Port	Type	Description
Input
input	(number \| null)[]	Any numeric series with matching timestamps.
Outputs
values	(number \| null)[]	Per-bar function result. Null for bars where the window extends before the first bar.
timestamps	number[]	Bar timestamps, aligned 1-to-1 with values.
func	WinFuncType	Active function name (mirrored for display).

Use Cases

Rolling highest high / lowest low (Donchian channel)

Apply func = max to the high series and func = min to the low series, both with period = 20. The result is a Donchian channel you can compare against close for breakout signals.

Adaptive volatility for position sizing

Use func = std (or volatility) with a 20-bar window on close returns. Feed the output into Position Sizing as a dynamic stop multiplier: when volatility expands, the stop widens automatically.

Trend detection via slope

Apply func = slope with period = 14 to a moving average series. When slope > 0 the trend is rising; when < 0 it is falling. Feed this into a Compare node to generate a binary trend direction flag.