RollingKurtosis Node
Rolling tail risk and outlier probability measurement
Overview
Rolling Kurtosis measures the frequency and magnitude of extreme events (fat tails) in rolling windows. Excess kurtosis > 0 = fatter tails (more crashes/spikes), excess kurtosis < 0 = thinner tails (smoother). Rising rolling kurtosis warns of increasing tail risk and potential black swan events.
Standard deviation misses tail risk. Kurtosis captures it. A strategy with Normal returns (K≈0) differs fundamentally from one with fat tails (K>3). Rising rolling kurtosis = widen stops, reduce leverage, hedge tail risk.
Formula & Calculation
Normal distribution: K = 3
K_excess > 0: Fat tails (crash risk)
K_excess < 0: Thin tails (smooth returns)
Parameters
| Parameter | Default | Description |
|---|---|---|
| lookback | 60 | Rolling window for kurtosis |
| excess | True | Use excess kurtosis (K-3) |
Common Use Cases
1. Tail Risk Detection
Rising kurtosis = tail risk rising. Widen stops from 2% to 3%. Add tail hedges (long VIX). Reduce leverage proportionally.
2. Leverage Adjustment
Leverage = Base / (1 + Kurtosis). High kurtosis periods = less leverage. Protect capital from outliers.
3. Regime Detection
High kurtosis periods = market stress. Switch to defensive strategies. Low kurtosis = opportunity for aggressive strategies.
4. Position Sizing
Position = Base × Normal_K / Current_K. High kurtosis trades = smaller positions (higher risk of tail events).
Advantages & Limitations
Advantages
- Detects tail risk early
- Simple to calculate
- Guides leverage decisions
- Anticipates black swans
Limitations
- Requires many samples
- Unstable in small windows
- Backward looking
- Doesn't predict severity