SELF-LEARNING PREDICTION MACHINE

Workshop Tutorial: Bitcoin Prediction, Backtesting, Correlation Analysis & Causation Studies

Read the Paper on blog.macropulze.com

Granger Causality Analysis - Statistical Relationship Testing

▶ A foreword of caution

Granger Causality Analysis

The Granger causality test determines whether one time series can predict another. This analysis examines whether Bitcoin, Gold, NASDAQ, and 10-Year Treasury yield have predictive relationships with each other over different time horizons.

Methodology

  • Granger Causality Test: Statistical test to determine if past values of X help predict Y
  • Null Hypothesis: X does not Granger-cause Y (p-value > 0.05)
  • Alternative Hypothesis: X does Granger-cause Y (p-value ≤ 0.05)
  • Data Processing: Log returns calculated for stationarity
  • Lag Selection: Optimal lags determined using AIC/BIC criteria
Analysis Period
30 Days
Data Frequency
Daily
Optimal Lags
2
Sample Size
30

Select Analysis Period

Granger Causality Test Results

Detailed results for each causality test. Lower p-values indicate stronger evidence of causality.

Loading Granger causality results...

Causality Network Graph

Interactive network showing directional Granger causality relationships between assets. Arrow thickness indicates significance strength, colors represent p-value levels.

Visualization Not Available

The causality network graph is optimized for larger screens. Please view on a desktop or tablet for the full interactive visualization.

p < 0.01 (Highly Significant)
p < 0.05 (Significant)
p < 0.10 (Marginal)
p ≥ 0.10 (Not Significant)

This visualization is best viewed on desktop screens

P-Value Heatmap Matrix

Color-coded matrix showing all causality relationships at once. Hover over cells for detailed information about each relationship.

Visualization Not Available

The p-value heatmap matrix is optimized for larger screens. Please view on a desktop or tablet for the full interactive visualization.

P-Value Scale:
0.00
1.00
Red = Highly Significant | Yellow = Significant | Gray = Not Significant

This visualization is best viewed on desktop screens

Bitcoin Predictive Relationship Analysis

Statistical analysis across multiple time horizons identifying leading indicators for Bitcoin price movements

Strongest Predictive Factor

NASDAQ Composite Index
Consistency
0/5 periods
Average p-value
0.4048

Most Consistent Predictor

Bitcoin → NASDAQ Causality
4/5 time periods significant
Minimum p-value: 0.0000

Weakest Predictive Factor

Gold Spot Price
Average p-value: 0.6426
Significant periods: 0/5

▶ Interpretation Guide (click to toggle)