Back to Docs
T
Traderoid

FractalDimension Node

Measures complexity and self-similarity in price patterns

StatisticalComplexityAdvanced

Overview

Fractal Dimension quantifies the complexity and self-similarity of price movements. Markets with high fractal dimension are more complex and chaotic. Markets with lower fractal dimension are more ordered and predictable. This metric helps assess whether markets are in regime where patterns will persist or break down.

Fractal analysis reveals that market structure appears similar across different timeframes - a property called self-similarity. Price movements show fractal-like patterns that can be exploited for regime detection and volatility prediction. High complexity often correlates with high volatility and lower predictability.

Formula & Calculation

Fractal Dimension (Box Counting)
D = log(N) / log(1/r)
Where N is number of bins needed to cover curve, r is scale
Interpretation
D = 1.0: Smooth line (perfect order)
D = 1.5: Random walk (moderate complexity)
D = 1.9: Very rough (high chaos)
D = 2.0: Space-filling (maximum disorder)

Parameters

ParameterDefaultDescription
lookback100-500Number of periods for calculation
bins20-50Resolution of grid for box counting

Common Use Cases

1. Market Complexity Assessment

High fractal dimension indicates complex market. Reduce position size and expect less predictable behavior.

2. Regime Detection

Track fractal dimension changes to identify shifts from trending to choppy markets or vice versa.

3. Volatility Correlation

Fractal dimension typically increases during high volatility periods. Use as early warning of regime change.

4. Pattern Recognition

Low complexity markets show more repeatable patterns. Increase strategy allocation when fractal dimension drops.

Advantages & Limitations

Advantages

  • Novel market complexity perspective
  • Detects regime changes early
  • Correlates with volatility
  • Works with non-stationary data
!

Limitations

  • Difficult interpretation for traders
  • Limited trading history validation
  • Computationally intensive
  • Parameter dependent results

Related Nodes