nyasekai

Graveyard / Calendar effects / US stocks

Trading the lunar cycle

Do stocks do better around the new moon?

Family
Calendar effects
Verdict
Dead
Cause
Overfit
Buried
9 October 2026

The idea

Studies reported a gap of several percent a year between new-moon and full-moon periods across dozens of countries. Hold SPY within about a week of each new moon and otherwise stay out.

How we tested it

DataSPY held on day t+1 when the close of day t falls within a quarter of a synodic month of a new moon. Variants: SPY, IWM and EFA together, and a tighter window of three days.
Period1995 to October 2026, split at June 2020.
Costs1 bp per side, 6.5 bp as a stress case.
HoldoutIn-sample and out-of-sample Sharpe.
ControlSPY buy-and-hold Sharpe 0.65. No random-entry control was run for this test.

Result

Primary, trades393
Sharpe full / in-sample / out-of-sample0.61 / 0.64 / 0.50
Max drawdown−37%
Three-index variant, out-of-sample Sharpe0.03
New-moon minus full-moon gap, 1995-2002+1,761 bp a year (t 1.2)
Same gap, 2003-14 / 2015-26+1,562 bp (t 1.4) / −716 bp (t −0.7)

Why it failed

Sharpe was below 0.8 with a 37% drawdown, and the gap reversed sign after 2015. The strategy is half-exposure SPY, so most of what it earns is market beta. No t-statistic in any period reached 2.

What would make it worth retesting

No. The proposed mechanism, investor mood, is a weak prior and the effect has not held up.

Test id: lunar-spy-event-v1. Past results of a backtest say nothing certain about the future.