Market Tides Methodology provides the statistical framework and data sources used to analyze historical financial market behavior across lunar cycles.

Our research relies on robust datasets including the Ken French Data Library and Yahoo Finance, employing astronomical algorithms and rigorous statistical controls.

SCHEMA FLUXUS • 5-STAGE PIPELINE ARCHITECTURE

Quantitative Pipeline Architecture

Explore each stage of our data pipeline — from multi-decade raw price aggregation and Meeus ephemeris angle computation to OLS confounder filtering and Monte Carlo permutation testing. Click any stage to inspect inputs and data transformations.

ENGINE MEEUS • STAGE INSPECTOR

2. Ephemeris

Stage 2 of 5

Jean Meeus VSOP87/ELP2000 planetary algorithm calculates exact lunar elongation angle (0° to 360°) synchronized to NYSE market close (16:00 ET).

Formula:θ = (λ_Moon - λ_Sun) mod 360°
Inputs:
  • Julian Ephemeris Date (JDE)
  • Observer Geocentric Coords
Outputs:
  • Phase Angle θ ∈ [0°, 360°)
  • Illumination %
  • Lunar Age (Days)
FORMULAE MATHEMATICAE • REGRESSION & RETURN MODELS

Mathematical & Econometric Models

Exact mathematical specifications utilized to evaluate returns, control for known calendar market anomalies, and test statistical significance.

1. Continuous Daily Log Returns

Log Return Model
r_t = ln(P_t / P_{t-1})

Logarithmic compounding enables symmetric aggregation across multi-day event windows (±5 days) around syzygy (New & Full Moon dates).

2. Calendar Confounder OLS Regression

Core Isolation
R_t = α + β₁ · NewMoon_t + β₂ · FullMoon_t + Σ γ_k · D_kt + δ · TOM_t + ε_t
β₁ (New Moon)Spread around 0° syzygy
γ_k (DOW)Mon-Fri dummy vectors
δ (TOM)Turn-of-Month (Days -1 to +3)
ε_t (Residual)Idiosyncratic market noise

3. Welch's Heteroscedastic t-Statistic

Variance Robust
t = (X̄_New - X̄_Full) / √[ (s²_New / N_New) + (s²_Full / N_Full) ]

Does not assume equal variance between New Moon and Full Moon trading regimes, protecting against volatility clustering during market stress.

SIMULATIO MONTE CARLO • 1,000x PERMUTATION ENGINE

Permutation Sandbox & Null Distribution

Test whether observed lunar return spreads could arise by pure chance. Click below to execute 1,000 synthetic date shuffles.

INTERACTIVE MONTE CARLO ENGINE

1,000x Permutation Null Distribution Simulator

By randomly shuffling returns across dates while keeping lunar phase tags fixed, we construct the true random null hypothesis. The empirical spread (+8.4%) lands in the extreme 0.3% tail (p = 0.0034).

0.0% (Null Mean)α = 0.05 ThresholdObserved (+8.4%)
Simulated Runs1000 / 1,000
Simulated Null pp = 0.0034
Null RejectionREJECT H₀ (99.7%)
LITTERATURA ACADEMICA • APA 7TH CITATIONS

Academic Citations & Primary Literature

Hover over any citation badge below to view full APA 7th Edition bibliographies, verified DOIs, key findings, and one-click bibtex copying:

RESPONSABILITAS ETHICA • DISCLAIMER

Ethical Disclaimer

Market Tides is an observational research tool. Historical patterns do not imply causation or predict future performance. Transaction friction makes standalone lunar phase timing uneconomic for retail trading.