Issues

• Schur likelihood paper incorrectly claims damping always lowers condition number

  microprediction/precise#125 · Filter by author microprediction opened on Oct 8, 2026

• Rank spectral calibration breaks equal-eigenvalue symmetry

  microprediction/precise#124 · Filter by author microprediction opened on Oct 7, 2026

• Spectral-calibration residual labels are not conditionally unbiased for covariance

  microprediction/precise#119 · Filter by author microprediction opened on Oct 5, 2026

• Literature map’s “ONLY robustified Cholesky-regression model” claim misses 2013–2014 work

  microprediction/precise#118 · Filter by author microprediction opened on Oct 4, 2026

• Two Sides overgeneralizes a scalar predictive γ* as the shared optimum for spatial and portfolio losses

  microprediction/precise#117 · Filter by author microprediction opened on Sep 28, 2026

• SchurConditional applies the OLS Schur identity to a ridge hedge, producing negative conditionals

  microprediction/precise#116 · Filter by author microprediction opened on Sep 28, 2026

• DCCCovariance's absolute variance floor breaks unit equivariance for mature low-volatility series

  microprediction/precise#115 · Filter by author microprediction opened on Sep 27, 2026

• GMVVariance scores a second moment, so return means can reverse covariance rankings

  microprediction/precise#114 · Filter by author microprediction opened on Sep 27, 2026

• DynamicUniverse makes a one-observation entrant look riskless beside mature assets

  microprediction/precise#113 · Filter by author microprediction opened on Sep 27, 2026

• TylerCovariance's ordinary mean gives one outlier hundreds of delayed shape updates

  microprediction/precise#112 · Filter by author microprediction opened on Sep 27, 2026

• HuberCovariance has unbounded outlier influence throughout burn-in

  microprediction/precise#111 · Filter by author microprediction opened on Sep 27, 2026

• Block assessors can reverse model rankings when asset columns are permuted

  microprediction/precise#110 · Filter by author microprediction opened on Sep 27, 2026

• geodesic_step retains an absolute 1e-12 floor, breaking scale equivariance and gamma=0

  microprediction/precise#108 · Filter by author microprediction opened on Sep 27, 2026

• set_params can silently desynchronize fitted estimator parameters from live state

  microprediction/precise#106 · Filter by author microprediction opened on Sep 26, 2026

• SchurConditional coupling reliability stays below one for deterministic correlations

  microprediction/precise#102 · Filter by author microprediction opened on Sep 26, 2026

• FixedUniverse's default ffill repeats missing returns and fabricates risk

  microprediction/precise#101 · Filter by author microprediction opened on Sep 26, 2026

• ConditionalCovariance startup floor makes correlation unit-dependent for hundreds of ticks

  microprediction/precise#100 · Filter by author microprediction opened on Sep 26, 2026

• Geodesic Schur interpolation changes the within-block risks it promises to preserve

  microprediction/precise#99 · Filter by author microprediction opened on Sep 26, 2026

• DiagonalCovariance burn-in is not translation invariant and fabricates variance

  microprediction/precise#98 · Filter by author microprediction opened on Sep 26, 2026

• DynamicUniverse missing-pair fill lets one asset's units rewrite unrelated risk

  microprediction/precise#97 · Filter by author microprediction opened on Sep 26, 2026

• Nonlinear shrinkage small-variable safeguard deletes a real direction and breaks precision

  microprediction/precise#96 · Filter by author microprediction opened on Sep 26, 2026

• LedoitWolf outlier cap is disabled for the first post-burn observation

  microprediction/precise#94 · Filter by author microprediction opened on Sep 25, 2026

• Pseudoinverse makes Huber and adaptive updates treat nullspace shocks as zero surprise

  microprediction/precise#93 · Filter by author microprediction opened on Sep 25, 2026

• OASCovariance assumes 1/r observations during burn-in and misstates mature ESS

  microprediction/precise#92 · Filter by author microprediction opened on Sep 25, 2026

• GMVVariance mishandles singular covariance via pseudoinverse normalization

  microprediction/precise#91 · Filter by author microprediction opened on Sep 25, 2026