Asset Correlations & Cointegration Guide (Asset Correlations)

1. Overview of Cointegration & Asset Correlations

Standard investors rely on Pearson correlation coefficients to assess the relationship between asset classes. However, correlation coefficients only trace short-term rate of changes. They cannot confirm long-term equilibrium and are highly vulnerable to spurious correlation errors.

This guide details the mathematical principles of Cointegration and how to leverage it for robust portfolio diversification.

---

2. Core Concepts & Statistical Metrics

① Pearson Correlation Coefficient

• Definition: Measures the linear relationship between two variables, ranging from -1.0 (perfect inverse relationship) to 1.0 (perfect positive relationship).
• Significance: Indicates whether daily price movements fluctuate in tandem under short-term sentiment cycles.

② Johansen Cointegration Test & p-value

• Definition: Evaluates whether a linear combination of non-stationary time series assets yields a stationary, long-term equilibrium spread.
• p-value Thresholds:
  • <= 0.05: Confirms a strong cointegrating relationship. Even if short-term spreads drift, the prices are bound by a structural connection that inevitably pulls them back to equilibrium.
  • > 0.10: No cointegration. Any perceived price alignment is temporary (spurious), and the asset prices can drift apart indefinitely.

---

3. Structural Dynamics: Correlation vs. Cointegration

• Spurious Alignments:
  • High Correlation, No Cointegration: Two assets move together briefly due to macro shocks, but lack structural connections (e.g. mutual supply loops, identical interest spreads). Arbitraging their spread carries immense risks of catastrophic divergence.
  • Low Correlation, High Cointegration: Daily price changes look random, but their spread never exceeds a constant envelope because of business structures (e.g. parent-subsidiary links, oil vs. jet fuel costs).

---

4. Practical Institutional Application

1. Detecting Decoupling Events:
   • If the p-value of Bitcoin and Nasdaq cointegration shifts from 0.02 to 0.15, the historical correlation has broken down. Valuation models must isolate their liquidity channels immediately.
2. Pairs Trading Safety Nets:
   • When a structurally cointegrated pair (e.g., TSMC and ASML) experiences extreme spread divergence, investors sell the overvalued asset and purchase the undervalued asset. Profits are collected as the spread mean-reverts back to statistical equilibrium.