Class VolatilityDecomposition

java.lang.Object
com.quantfinlib.volatility.VolatilityDecomposition
public final class VolatilityDecomposition extends Object
Systematic vs IDIOSYNCRATIC volatility — the decomposition behind "how much of this stock's risk is the market, and how much is the company?" A single-factor (CAPM-style) regression of asset returns on market returns splits total variance EXACTLY: 
  Var(asset) = β²·Var(market)  +  Var(residual)
               └── systematic ──┘   └ idiosyncratic ┘
with β = Cov(asset, market)/Var(market) — the split is exact (not approximate) because OLS residuals are uncorrelated with the regressor by construction. The two halves behave differently and deserve different treatment: systematic volatility (rates, inflation, macro shocks) cannot be diversified away and is hedgeable with index instruments; idiosyncratic volatility (earnings, management, product launches) diversifies across names and is exactly what single-name hedges, pairs trades and the CRB's per-symbol factors carry.

R² is the systematic SHARE — a utility at 0.7 is mostly a market proxy; a biotech at 0.05 is mostly its own story. All variances are per-period (annualize with ×periodsPerYear, vols with the square root — the record has helpers). Sample moments (n−1) throughout. Static, deterministic, research lane; pairs with risk.RiskMetrics.beta (same β, cross-checked in tests) and alpha.PortfolioConstruction's beta-neutralization.

  • Method Details

    • decompose

      public static VolatilityDecomposition.Decomposition decompose(double[] assetReturns, double[] marketReturns)
      Decomposes an asset's variance against a market/benchmark series.
      Parameters:
      assetReturns - per-period returns, ≥ 30 finite observations
      marketReturns - aligned benchmark returns (must carry variance — a flat benchmark decomposes nothing)