Struct FactorModel 

pub struct FactorModel;

Fama-French style multi-factor OLS regression engine.

Algorithm

Given T observations and K factors, form the T × (K+1) design matrix X (first column all-ones for the intercept) and solve the normal equations:

β̂ = (X'X)⁻¹ X'y

Matrix inversion is done analytically for 2×2 and 3×3 systems and via Gaussian elimination with partial pivoting for larger systems.

Implementations

impl FactorModel

pub fn fit( asset: &str, asset_returns: &[f64], factors: &[Factor], ) -> FactorExposure

Fit a factor model to asset_returns using the provided factors.

Panics

Does not panic; returns a zero-exposure result if the system is singular or if the return series have mismatched lengths.

pub fn decompose( exposure: &FactorExposure, factor_cov: &CovarianceMatrix, factor_names: &[String], ) -> VarianceDecomposition

Decompose total variance into systematic and idiosyncratic components.

factor_contributions[i] = β_i² σ_i² + 2 Σ_{j>i} β_i β_j cov(i,j)