Monte Carlo test of harmonic phase recovery for a synthetic light curve with a true secondary minimum (Δϕ = 180°). Top: Example realisation showing the noiseless model (black dashed), noisy sampled data with seasonal cadence (blue points), and the best-fitting two-harmonic model recovered at the known fundamental and harmonic frequencies (red). Bottom: Distribution of recovered relative phases Δϕ = ϕ₂ − 2ϕ₁ over 2 × 10⁴ Monte Carlo realisations. The shaded region indicates a ±10° window around Δϕ = 0, corresponding to the secondary maxima regime. No significant population of noise-driven phase flips from Δϕ = 180° to Δϕ ≈ 0 is observed, demonstrating that such misclassification is negligible at the signal-to-noise levels considered.