Back to Docs
T
Traderoid

HurstExponent Node

Measures trending vs mean-reverting behavior

StatisticalTrendRegime

Overview

Hurst Exponent (H) characterizes whether a time series is trending, mean-reverting, or random. H=0.5 indicates random walk. H>0.5 indicates persistent behavior (trending). H<0.5 indicates anti-persistent behavior (mean reverting). This single number tells you whether to use momentum or mean reversion strategies.

The Hurst Exponent is derived from rescaled range (R/S) analysis and quantifies long-range dependencies in data. It's more reliable than simple autocorrelation for detecting regime changes and helps traders adapt their strategies dynamically to current market conditions.

Formula & Calculation

Rescaled Range (R/S)
H = log(R/S) / log(N)
Where R=Range, S=StdDev, N=Period
Interpretation
H = 0.5: Random walk (50%% = random)
H = 0.6+: Trending market (use momentum)
H = 0.4-: Mean reverting (use reversion)
H > 0.7: Strong trending (bubble/crash risk)

Parameters

ParameterDefaultDescription
lookback100-500Periods for Hurst calculation
methodRSCalculation method (RS or DFA)

Common Use Cases

1. Strategy Selection

H>0.55: Use momentum/trend-following strategies. H<0.45: Use mean reversion. H near 0.5: Avoid both or use oscillators.

2. Market Regime Detection

Track Hurst monthly to detect regime shifts. Rising Hurst signals transition from mean reverting to trending markets.

3. ATR Adjustment

Increase position size when Hurst is high (trending), reduce when low (mean reverting). Match position size to market regime.

4. Divergence Detection

When price makes new high but Hurst falls, suggests ending trend. Vice versa for bottoms. Potential reversal warning.

Advantages & Limitations

Advantages

  • Distinguishes trending from mean reverting
  • Guides strategy selection
  • Quantitative regime detection
  • Works well with other indicators
!

Limitations

  • Ambiguous at 0.5 (hard to interpret)
  • Requires large sample (100+ bars min)
  • Unstable over time
  • Different methods give different results

Related Nodes