Lunar Phase Cycle Indicator: Does It Actually Work?

Tonight, Wednesday, August 12, 2026, the Moon passes directly in front of the Sun. Around 8:30 p.m., a 294-kilometer band of shadow sweeps across northern Spain — Oviedo, Gijón, Santander, Bilbao, Burgos, Zaragoza. Mainland Spain hasn’t seen this since 1905: 121 years.

And like every solar eclipse, this one falls on a very precise day of the lunar calendar: the new moon. A geometric constraint, not a coincidence.

In other words: tonight, a certain indicator turns green. It’s called the Lunar Phase Cycle, it really exists, and it drops onto a chart in two clicks. So we did what we always do here with an indicator: we tested it. Across 131 lunations, on two assets.

The key takeaway:

• The Lunar Phase Cycle doesn’t calculate anything from price: it lays an astronomical calendar over your chart.
• The classic rule — buy at the new moon, sell at the full moon — returns +1,216 % versus +997 % for the inverse half since 2016. In other words, next to nothing.
• On the S&P 500, the signal is flat-out on the wrong side: +60 % versus +141 % for the half you’re supposed to avoid.
• In both cases, doing absolutely nothing crushes both halves.

What Is the Lunar Phase Cycle?

It’s an indicator you add to your chart like any other, available in community libraries — TradingView chief among them — under various names: Lunar Phases, Moon Phases, Lunar Phase Cycle.

It displays a marker at every new moon and every full moon, often with a colored background alternating between the two, sometimes a sine wave representing the illuminated fraction of the disk. The most common rule of thumb fits in one sentence: buy at the new moon, sell at the full moon.

And there’s one thing to understand right away: this indicator never looks at price. The RSI reads closing prices, the Volume Profile reads traded volume, the CVD reads order aggressiveness. The Lunar Phase Cycle, on the other hand, reads an ephemeris.

It would produce exactly the same signals on Bitcoin, on wheat, or on a temperature chart. It’s a calendar disguised as an indicator — which already says a lot about what you can expect from it.

What It Plots, in Two Minutes

The Moon doesn’t emit light, it reflects it. The half facing the Sun is always lit; what changes over the course of the month is how much of that half we can see from Earth.

The full cycle is called a lunation: 29.530589 days on average, or 12.37 lunations a year. When the Moon slips between Earth and the Sun, it shows us its dark side — the new moon, and the only moment a solar eclipse can happen. Remember that decimal figure, it comes back further down.

We Tested It on 131 Lunations

The protocol, announced before the results — the least we can do:

• Assets: Bitcoin and the S&P 500, daily closes, from January 1, 2016 to August 6, 2026 — 131 complete lunations.
• Phase dates: calculated with the Meeus method, accurate to the minute — not a simple multiple of 29.53 days, which drifts by fourteen hours.
• Rule: entry at the close following the new moon, exit at the close following the full moon. Then we also measure the inverse half, the one you’re supposed to avoid.

On Bitcoin, the rule racks up +1,216 %. Impressive — until you look at the inverse half, the one you’re supposed to spend out of the market: +997 %. That’s 3.0 % per half-cycle versus 2.8 %. Two-tenths of a point apart, on half-cycles that swing by dozens of points. That’s noise.

On the S&P 500, it’s more embarrassing: the “new → full” half returns +60 %, the one you’re supposed to avoid returns +141 %. The signal isn’t just weak, it points in the wrong direction.

And the number that ends the debate: over the same period, doing nothing — buying and holding — returns +14,696 % on Bitcoin and +284 % on the S&P 500. Both lunar halves lose to inaction, mechanically, since each one only spends half its time in position.

In fairness, since that’s the whole point: this test doesn’t prove anything either. Two assets, one period, one entry convention, no fees, no slippage. It illustrates, it doesn’t demonstrate. If the result had been spectacular in the other direction, it would have deserved exactly the same skepticism.

And Yet, Serious Studies Do Find an Effect

That’s true, and it’s what makes the topic fun.

In 2001, Ilia Dichev (Michigan) and Troy Janes (SUNY Buffalo) published Lunar Cycle Effects in Stock Returns: returns over the fifteen days surrounding a new moon are worth roughly double those surrounding a full moon, across a century of U.S. indices and twenty-four other countries. Two caveats, though: the paper ended up in the Journal of Private Equity, a practitioner journal rather than a top-tier academic one; and the annualized gap attributed to it ranges from 4.8 % to 9.4 % depending on who’s citing it. When the same paper produces four different figures across four citations, that’s already telling you something.

The strongest contribution comes from Kathy Yuan (London School of Economics), Lu Zheng, and Qiaoqiao Zhu, in 2006, in the Journal of Empirical Finance — a genuine top-tier reference this time. Across 48 countries, they measure lower returns around the full moon, with a gap of 3 to 5 % a year. And they check that it isn’t something else in disguise: the effect survives controls for the January effect, the day-of-the-week effect, the turn-of-the-month effect, and public holidays.

Their hypothesis is psychological, not gravitational: the full moon would degrade sleep and mood, nudging investors toward a slightly pessimistic bias. That’s consistent with the effect being stronger for small caps and in countries with high individual stock ownership. Nothing esoteric about it — it’s market psychology, a field we take very seriously.

Except the research doesn’t stop there.

In 2007, Anthony Herbst revisits the Dow Jones in the Journal of Bioeconomics and concludes: no consistent, predictable lunar influence, neither on returns nor on volatility. In 2011, Keef and Khaled test 62 indices over twenty years in that same Journal of Empirical Finance: the new-moon effect survives, the full-moon effect disappears. Half of the original result doesn’t replicate.

And crypto? A Turkish team published a Bitcoin-focused study with Springer in 2023, backed by a McNemar test. The conclusion, no punches pulled: no statistically significant impact. You can indeed find papers claiming lunar strategies with over 30 % annualized returns on BTC — they come out of journals whose names show up on predatory-publisher watchlists, the kind that will publish almost anything for a fee. Same question, two opposite answers, two utterly incomparable standards of rigor.

Why You Always End Up “Finding” Something

Here’s the only genuinely useful part of this article.

Take a significance threshold of 5 %: “a 5 % chance of seeing this by pure luck.” Now test 20 random ideas: the odds that at least one looks significant while being worthless climb to about 64 %. At 100 tests, you’re guaranteed to find one.

Worse: in 2011, Simmons, Nelson, and Simonsohn showed that combining just four small methodological liberties — picking your variables after the fact, dropping two outliers, stopping data collection at a convenient point, testing two measures and keeping the better one — pushes the false-positive rate from 5 % to over 60 %. Without consciously cheating a single time.

Finance knows this problem very well. In 2016, Campbell Harvey (Duke) and his co-authors counted 316 factors in the Review of Financial Studies, published between 1967 and 2014, every one of them supposedly explaining stock returns. Three hundred and sixteen. Their conclusion: at this level of collective data-mining, the usual threshold (t-stat above 2) stops meaning anything, and the bar needs to move to t-stat above 3.

And what happens to these anomalies once published? McLean and Pontiff, in the Journal of Finance, tracked 97 predictive variables: their returns drop by 26 % out of sample and by 58 % after publication. Anomalies wear out the moment you use them.

The best antidote is still visual. Tyler Vigen built an entire site of absurd, automatically generated correlations: the number of Nicolas Cage films per year correlates at 0.666 with swimming-pool drownings in the United States; per-capita cheese consumption correlates at 0.947 with the number of people who died tangled in their bedsheets. Coefficients any strategy would dream of.

Nobody’s going to trade cheese. But “the Moon” sounds mysterious enough to believe in — and that’s exactly the trap. If you’re interested in the probability angle, our article on poker and trading digs into the same idea with cards instead of celestial bodies.

The Design Flaw: Three Clocks That Never Realign

One purely arithmetic obstacle remains. A lunation lasts 29.53 days, the average calendar month 30.44: a 0.91-day gap per month, which accumulates. A full moon therefore drifts across the entire calendar over the course of the year.

Worse: 29.53 divided by 7 gives 4.2186 weeks. Not a whole number. A lunar signal therefore wanders across every day of the week — weekends included, when equity markets are closed. Hence a delicious asymmetry between our two assets: Bitcoin trades seven days a week and catches every signal, while the S&P 500 misses two out of seven.

The Indicators That Actually Measure Something

If you came looking for a cycle indicator, here are some real ones. What they have in common: they describe what participants are doing right now, not the position of a natural satellite.

On the positioning and flow side, the data that’s hardest to fake:

• The Cumulative Volume Delta measures whether buyers or sellers are chasing price aggressively.
• Open Interest tells you how many positions are genuinely open: a rally without OI doesn’t mean the same thing as a funded rally.
• The funding rate reveals the cost of staying positioned, and so the excess complacency on one side of the market.
• The COT Reports detail the weekly positioning of major players, and order flow gives you the same information down to the second.

On the price structure side: the Volume Profile and the Market Profile show where the market has actually traded, the VWAP serves as a reference for thousands of institutional executions, the Average True Range calibrates a stop loss instead of placing it at random, and the Chop Index sorts trend from range. For a quick way in, our pick of the Captain’s three favorite indicators does the sorting for you — the classics still hold their own too: RSI, Bollinger Bands, moving averages.

That leaves the data platforms, the ones that produce these numbers. An indicator is never better than the source feeding it:

• Our comparison of five order-flow analysis tools, the best place to start if you don’t know where to begin.
• Coinalyze for aggregated open interest, funding, and liquidations; Exocharts and Market Monkey for footprint charts; The Kingfisher for liquidation zones.
• The CME Group for official futures data, and TradingView for the chart itself — the very place you’ll find the Lunar Phase Cycle, if curiosity ever gets the better of you.
• Forex Factory and CoinMarketCal for the economic calendar — the only calendar that actually moves markets.

You’ll notice that correlation between assets, on the other hand, can actually be measured and documented — exactly what we do in our analysis of the correlation between the S&P 500 and crypto. A correlation explained by identifiable capital flows has nothing to do with a correlation explained by nothing.

And if it’s the word “cycle” that drew you in: markets really do have four phases — accumulation, expansion, distribution, decline — but they’re read on the chart instead of checked off on a calendar. That’s the subject of our guide to the four phases of a market cycle, its take on the crypto cycle, and the Wyckoff method. Richard Wyckoff had no ephemerides, only a tape of quotes.

Want to Test It Anyway? Five Habits

Go for it, honestly. It’s a good excuse to learn how to test something. Five habits are all it takes, and they apply to every indicator:

1. Set the hypothesis BEFORE looking at the data. “Buy at the new moon, sell at the full moon” is a hypothesis. “Look for what works around the Moon” is not.
2. Always measure the inverse half. The single most worthwhile test there is. If the opposite rule performs almost as well, you haven’t found a signal: you’ve found the market.
3. Compare it to “doing nothing.” A rule that gains 1,200 % but still trails buy and hold isn’t a strategy, it’s a detour.
4. Hold out a sample. Fit the rule on one half of the history, check it on the other. If it doesn’t survive, it never existed.
5. Count your attempts. Twelve variants before finding the one that works? Then it isn’t the rule that’s working, it’s your patience.

And write them down somewhere, these attempts — failures included. That’s exactly what a trading journal is for, and ours is free. The rest of the serious toolkit is there too: strategy, trading plan, risk management. Nothing lunar about any of it.

Tonight, Then

Around 8:30 p.m. local time, totality will sweep across Asturias, Cantabria, the Basque Country, Navarre, and Aragon. Oviedo gets the longest duration on the peninsula: 1 minute and 49 seconds of nighttime in the middle of the evening. Don’t fall for the common assumption — Madrid and Barcelona are not in the path of totality — they’ll see a partial eclipse, a very deep one, but partial all the same. From Paris, it’ll be 92.1 % of the disk covered around 8:17 p.m.

So there it is. The Lunar Phase Cycle predicts nothing, measures nothing, and even gets the direction wrong half the time. It’s probably the most useless indicator ever installed on a chart — and yet it’s magnificent.

Because it has one merit no other indicator has: it forces you to look up from the screen. Tonight, a few million people will watch exactly the same thing at exactly the same moment, without wondering for a single second whether it’s tradable. Certified glasses on, and enjoy the view.

Tomorrow, the markets will reopen exactly as they would have without an eclipse. And the volume delta, for its part, will still be there.

Frequently Asked Questions

What Is the Lunar Phase Cycle Indicator?

It’s a chart indicator that displays new moon and full moon dates over the price candles, often with a colored background alternating between the two phases. Its distinctive trait: it doesn’t calculate anything from the market, it overlays an astronomical calendar. It would produce the same markers on any asset. You’ll find it in community indicator libraries, notably on TradingView.

Does the “Buy at the New Moon, Sell at the Full Moon” Rule Work?

Not on our measurements. From January 1, 2016 to August 6, 2026, or 131 lunations, it racks up +1,216 % on Bitcoin against +997 % for the inverse half: a 0.2-point gap per half-cycle, indistinguishable from noise. On the S&P 500 it returns +60 % against +141 % for the half you’re supposed to avoid — the signal is inverted. In both cases, holding and doing nothing does noticeably better.

Does the Lunar Cycle Have a Proven Effect on Markets?

No, not in the sense of an established effect. Two studies find a gap between new and full moon (Dichev & Janes in 2001, Yuan, Zheng & Zhu in 2006 across 48 countries, at 3 to 5 % a year). But Herbst finds no consistent effect on the Dow Jones in 2007, and Keef & Khaled only validate half the original result in 2011, across 62 indices. An effect that half the literature can’t replicate can’t serve as the basis for a strategy.

How Long Does a Lunar Cycle Last?

The synodic month — from one new moon to the next — lasts 29.530589 days on average, or 29 days, 12 hours, and 44 minutes, give or take a few hours depending on the Moon’s position along its elliptical orbit. That’s 12.37 lunations a year, which is why the phases keep shifting through the calendar and regularly fall on weekends.

Does a Solar Eclipse Happen at a Full Moon or a New Moon?

Always at a new moon: a solar eclipse requires the Moon to sit between Earth and the Sun, which is the very definition of a new moon. The lunar eclipse, on the other hand, corresponds to the full moon. Not every new moon produces an eclipse, though, because the Moon’s orbit is tilted about 5 degrees relative to the plane of Earth’s orbit.

Does the Moon Influence the Price of Bitcoin?

The most serious academic study on the subject, published with Springer in 2023, applies a McNemar test to Bitcoin’s opening and closing prices at the boundaries of lunar cycles and finds no statistically significant impact. Our own measurements across 131 lunations point the same way. Claims to the contrary come from publications of dubious editorial credibility.

What Cycle Indicators Should You Use Instead?

The ones that measure what participants are actually doing: the Cumulative Volume Delta for buying or selling aggressiveness, Open Interest for open positions, the funding rate for the cost of positioning, the Volume Profile for trading zones, and the COT Reports for the positioning of major players. For reading market phases themselves, the Wyckoff method and the classic four-phase breakdown remain the go-to references.

Why Do So Many Indicators Seem to “Work” in Backtests?

Because testing lots of ideas mechanically produces positive results by chance. With a 5 % significance threshold, testing 20 independent hypotheses gives roughly a 64 % chance that at least one looks significant while being worthless. Academic finance has counted 316 published factors supposedly explaining returns; Campbell Harvey recommends, for this very reason, raising the usual validation threshold from 2 to 3 standard deviations.