Quantitative Research

·

May 2026

Full Moon Effects on
S&P 500 Returns

A backtest of 309 full moon events from 2000 through 2024 against S&P 500 performance across five holding periods — with rigorous multiple-testing correction applied.

☽

309 events · 2000–2024 · 5 holding periods · Bonferroni corrected

Executive Summary

The analysis covers 309 full moon occurrences from 2000 through 2024 — approximately 25 years of lunar cycles tested against S&P 500 performance. Five holding periods were examined (1, 3, 7, 14, and 30 days), requiring a Bonferroni-corrected significance threshold of 0.010 rather than the standard 0.05.

No statistically significant lunar effect was detected. None of the observed patterns survive multiple-testing correction, making any findings suggestive at best. The one-day forward returns averaged exactly 0.00% with only 38.8% winning sessions. The strongest result — the 14-day window — showed 0.44% average returns and a 63.1% win rate, but produced a p-value of 0.023, which fails the Bonferroni threshold.

Data & Methodology

Full moon dates were sourced for each calendar month from January 2000 through December 2024. S&P 500 returns were calculated from the first trading session on or after each full moon date across five forward-looking windows: 1, 3, 7, 14, and 30 calendar days. Where full moons fell on non-trading days, the next open session was used as the entry point.

A baseline return was computed for each matching holding period using all trading days in the same calendar sample, to isolate any excess return attributable to lunar timing versus the prevailing market trend. Because five separate holding periods were tested simultaneously, Bonferroni correction was applied: the significance threshold was divided by five, yielding a corrected threshold of 0.010. Any p-value above 0.010 cannot be considered statistically significant given the multiple-testing context.

Additional context was drawn from SPY ETF data covering March 2009 through March 2025 — a period starting at $51.77 and reaching $554.51, representing a 971.1% gain. This bull market backdrop is a critical confound: the lunar analysis captures predominantly rising market conditions where any behavioral effect must compete against powerful fundamental tailwinds.

Results by Holding Period

1-Day

Avg Return

0.00%

Win Rate

38.8%

Excess

+0.07%

P-Value

0.949

Near-zero return, below-average win rate

3-Day

Avg Return

0.04%

Win Rate

54.4%

Excess

—

P-Value

—

Modest improvement in win rate

7-Day

Avg Return

0.06%

Win Rate

53.4%

Excess

—

P-Value

—

No meaningful edge over baseline

14-DayStrongest Signal

Avg Return

0.44%

Win Rate

63.1%

Excess

+0.12%

P-Value

0.023

Strongest result — fails Bonferroni threshold of 0.010

30-Day

Avg Return

0.54%

Win Rate

61.2%

Excess

0.00%

P-Value

—

Any apparent edge dissolves entirely

Bonferroni-corrected significance threshold: 0.010 · None of the above patterns survive correction

What the Data Shows

The comprehensive backtest reveals no statistically significant lunar effect on S&P 500 returns around full moons. The one-day forward returns averaged exactly 0.00% — producing a marginal 0.07% excess return versus the baseline that fails significance testing with a p-value of 0.949. The below-average win rate of 38.8% is directionally contrary to any bullish lunar hypothesis.

The pattern does not improve meaningfully over longer horizons. Three-day returns averaged 0.04% with 54.4% wins. Seven-day returns averaged 0.06% with 53.4% wins. Both figures are consistent with random market noise across a predominantly rising sample period.

The 14-Day Window — Closest to Signal

The fourteen-day window presents the most intriguing result in the dataset. At 0.44% average returns, a 63.1% win rate, and a modest 0.12% excess return versus baseline, it produces a p-value of 0.023 — which would appear significant under standard single-test methodology.

It does not survive Bonferroni correction. Testing five holding periods simultaneously means the threshold for significance is 0.010. A p-value of 0.023 reflects roughly one-in-forty odds of observing this result by chance — meaningful for a single test, but insufficient when five tests are run at once, because the probability of at least one spurious "hit" across all five rises considerably.

The thirty-day window reinforces the conclusion: 0.54% average returns and 61.2% wins, but zero excess return versus baseline. Any appearance of outperformance over two weeks dissolves entirely into the general market trend over a month. This is the fingerprint of a non-effect: signal decaying toward the mean, not persisting as genuine alpha would.

14-day avg return

0.44%

Win rate

63.1%

P-value

0.023

Bonferroni threshold

0.010

Two Theoretical Frameworks

The lunar market hypothesis operates under two distinct theoretical frameworks that require separate evaluation. Each makes different predictions and each is tested differently by the backtest data.

Mechanistic

Direct Gravitational / Electromagnetic Influence

Refuted by data

The stronger claim: lunar phases exert independent pressure on investor psychology through gravitational or electromagnetic means, creating measurable price effects. Under this reading, fundamental drivers — earnings, Fed policy, geopolitical events — are genuine confounding factors competing for causal attribution. The backtest verdict is clear: with 309 signals across 25 years, no persistent outperformance survives after controlling for baseline market behavior.

Archetypal

Lunar Phases as Structural Timing Markers

Inconclusive

The subtler claim: full moons mark the timing of structural market ruptures rather than causing independent price movements. This framework argues that monetary regime changes, financial crises, and policy pivots cluster near lunar events as expressions of an underlying archetypal pattern. Under this reading, the excess-return test becomes a measure of whether full moons coincide with unusual market behavior relative to random timing — and the near-zero excess returns across all windows suggest they do not.

What This Means Now

The massive 2009–2025 bull market forms the dominant backdrop for this analysis. Starting at $51.77 and reaching $554.51, SPY returned 971.1% over the sample period — with a peak of $606.11 and a financial crisis trough of $49.20. Any lunar trading strategy operating in this environment was competing against powerful fundamental tailwinds that dwarfed any behavioral effect.

Recent performance during 2024–2025 exemplifies the inconsistency that undermines lunar trading strategies. Across recent full moons, the record documents mixed results: four gap-up openings, five bullish sessions, two bearish sessions, and one sharp decline. This erratic pattern aligns precisely with the backtest findings — no reliable directional bias, just random variation around a rising market trend.

The academic foundation traces to Dichev and Janes (2001), who documented that returns in the 15 days around new moons approximately doubled those around full moons across major U.S. indexes over 100+ years. For the S&P 500 specifically from 1928–2000, mean daily returns during new moons were 0.046% versus 0.024% during full moons. But the large standard deviation of daily returns around 1.1% rendered these differences statistically insignificant at individual index levels — a pattern the current backtest continues to replicate.

Actionable Implications

Given the absence of statistically significant lunar effects across 309 events, any trading strategy based on full moon timing requires the following constraints:

01

Position sizing for any lunar-based trade should not exceed 1–2% of portfolio allocation, reflecting the high probability that observed patterns represent statistical noise rather than genuine market inefficiency.

02

The 14-day window (0.44% average return, 63.1% win rate) provides the strongest historical precedent but failed significance testing. Treat it as a weak directional tilt, never a primary signal.

03

Key conditions that might amplify behavioral lunar effects: VIX above 25, unusual retail trading volume, major Fed policy uncertainty. Under normal conditions, ignore lunar timing in favor of earnings calendars and economic data.

04

Stop-loss levels should be set on technical and fundamental grounds — not on lunar cycle expectations. The 1.1% daily volatility backdrop means any lunar signal can be overwhelmed by a single news event.

05

Close any lunar-based position within two weeks. The 30-day zero excess return confirms that historical patterns provide no directional edge past the two-week mark.

A Note on Methodology

This analysis examines correlations between lunar cycles and market behavior with rigorous statistical controls, including Bonferroni correction for multiple comparisons. We make no claims about causation. The honest conclusion from 25 years of data is that full moon timing offers no reliable trading advantage in the S&P 500. While academic research documents lunar effects over century-long periods, the practical trading value remains questionable due to high volatility, inconsistent patterns, and modest effect sizes that disappear under rigorous statistical testing. The sample size of 309 events is statistically meaningful for detecting persistent patterns — the absence of a finding here is itself informative.

Disclaimer — This analysis is presented for educational and informational purposes only. It does not constitute investment advice, a solicitation to buy or sell any financial instrument, or a recommendation regarding any specific market action. The author is not a registered investment advisor. Past patterns do not guarantee future results. Statistical analysis involves inherent uncertainty; findings should not be acted upon without independent verification.

← Back to Market Astrology

Stay informed

Join Early Access