Docs MathStatistics

Statistics

34 nodes — Rolling-window statistical measures and risk metrics

Statistics nodes compute measures over a configurable rolling window: central tendency, spread, distribution shape, trend metrics, and risk ratios. Most nodes default to a 20-bar window and output null until the window warms up.

OLS slope over first-differences — measures the rate of change of momentum.

0-based index of the highest value in the window. Detects swing highs and peak age.

0-based index of the lowest value in the window. Detects swing lows and trough age.

Autocorrelation

Lag-1 Pearson correlation — distinguishes trending from mean-reverting regimes.

Arithmetic mean (SMA) over the window. Foundation for deviation and Z-score pipelines.

Number of non-null values in the window. Essential for data density validation.

Running total from bar zero. Converts returns into equity curves or cumulative volume.

Coefficient of Variation

Std / |mean| × 100 — normalised volatility for cross-asset comparison.

Shannon entropy via uniform binning — high entropy means noise; low means trending.

Sum of positive first-differences — total upward movement within the window.

OLS y-intercept of the rolling regression. Anchors the trend line at its start.

Q3 − Q1 — outlier-robust spread measure, insensitive to fat-tail price spikes.

Excess kurtosis — positive values indicate fat tails and elevated tail-risk.

OLS slope in price-units per bar. The canonical trend-velocity indicator.

Sum of absolute negative first-differences — total downward movement (positive output).

Mean absolute deviation — the spread measure used in CCI, robust to outliers.

Highest value in the window (N-bar high). Core component of Donchian channels.

Largest peak-to-trough decline as a fraction of the peak within the window.

Middle sorted value — a robust central tendency measure unaffected by outliers.

Lowest value in the window (N-bar low). Core component of Donchian channels.

(last − first) / |first| — the window's rate-of-return. Classic momentum factor.

Q1 (First Quartile)

25th percentile of the window. Defines the lower bound of the central distribution.

Q3 (Third Quartile)

75th percentile of the window. Defines the upper bound of the central distribution.

OLS coefficient of determination — measures how linearly clean the trend is (0–1).

Max − min over the window — simplest volatility proxy and Donchian channel width.

√(mean(x²)) — measures signal energy or amplitude regardless of sign.

Mean / std — rolling risk-adjusted return proxy assuming zero risk-free rate.

3rd standardised moment — negative skew warns of asymmetric downside tail risk.

OLS slope over the window. Positive → uptrend; negative → downtrend; zero → flat.

Mean / downside-std — penalises only negative volatility, unlike Sharpe.

Std Error Bands

OLS fitted ± multiplier × std error. Produces upper, middle, and lower band series.

Windowed · 3 outputs

Total of non-null values in the sliding window. Complements CumSum for fixed-period sums.

Population variance Σ(x−μ)²/n. Apply Sqrt for standard deviation and Bollinger Bands.

Std of period-over-period returns — realised volatility for vol-targeting strategies.