Guide

Golden cross and death cross in crypto: what they are, and what a test showed

A golden cross is the 50-day moving average of price rising above the 200-day moving average. A death cross is the 50-day falling below it. Traders read the first as the start of an uptrend and the second as the start of a downtrend. Held as a rule on the 20 most-traded crypto assets from 2019 to 2026, long from each golden cross to the next death cross, it returned 14.4% a year with a worst fall of 72.4%. Holding the same basket throughout returned 38.5% a year with a worst fall of 85.1%. Those figures come from one test of one market over one period, and this page sets out what they do and do not show.

What a golden cross and a death cross are

Both describe the same pair of lines. A moving average is the mean of the last N daily closes, recalculated each day. The 50-day average follows price fairly closely; the 200-day average follows it slowly. When the faster line climbs through the slower one, that is a golden cross. When it drops back through, that is a death cross. Between the two, the 50-day average sits above the 200-day, and that state, not the day of the cross, is what a rule holds.

The lengths are a convention, not a law. Fifty and two hundred days are the pair most people mean, usually with simple averages. Some use exponential averages or other pairs. The test described here used simple averages at 50 and 200 days and nothing else.

Why the signal arrives late

A 200-day average moves slowly and a 50-day average needs weeks to turn, so the two cross well after price itself has changed direction. That is the design: the delay filters out short moves, and it is why the crossover trades so rarely. It is also the cost. By the day a death cross prints, price has already moved some distance down from its high, and by the day a golden cross prints, some distance up from its low. In a market that moves as fast as crypto, that distance can be large.

What it did in the test

14.4%
a year, compounded
72.4%
worst fall from a peak
180
round trips, cross to cross
72%
of round trips lost money
130 days
the average trip

Each round trip is one golden cross followed by one death cross on one asset. There were 180 of them in the basket, about 1.2 per asset per year. 72% lost money. The average losing trip lost 27% and the average winning trip made 164%. Almost none were quick: 2% were lost within ten days.

Rule, long or cashRound tripsReturn per yearWorst fallLongest under waterTrips that lost
Golden cross to death cross (50/200)18014.4%72.4%1,980 days72%
Close above its 200-day average92613.9%76.2%1,980 days88%
Close above its 50-day average1,83337.6%48.8%1,185 days79%
Supertrend(10, 3)68242.5%46.1%922 days58%
Hold the basket—38.5%85.1%1,980 days—

The same data, basket, costs and execution for every row. From the pre-registered comparison of Supertrend with a moving average, run 2026-11-05, where these rows are a disclosure and no row is offered as the one to use.

How to read those numbers

  • A disclosure, not a tested claim. The comparison these figures come from registered one question in advance: Supertrend against one moving average. The crossover was run beside it and shown in full so that nothing was left out. No test of whether its figures differ from the others by more than chance was run on it.
  • A basket, not Bitcoin. The test holds the 20 most-traded crypto assets, chosen again each quarter. The golden cross is mostly discussed for Bitcoin alone, which this site has not measured.
  • One period. 2019-01-01 to 2026-10-09. A different stretch of history would give different figures, and a slow rule has few trips to judge it by.
  • Long or cash, with costs. The rule holds from a golden cross to the next death cross and sits in cash otherwise. It never sells short. The cross is read on the completed daily candle and acted on at the next open, at 0.30% a side.
  • Simple averages at 50 and 200 days. Exponential averages, other pairs of lengths and weekly candles were not tested.

What the figures do say is narrow and plain: on this basket over these years, the crossover rule still fell 72.4% at its worst, against 85.1% for holding, and it returned less than holding did. They describe the past. They are not a forecast.

The 200-day moving average on its own

A simpler rule uses one line instead of two: hold while the daily close is above its 200-day average. In the same test it made 926 round trips, returned 13.9% a year and had a worst fall of 76.2%. 70% of its trips were lost within ten days, because price that sits near a single average steps over it and back. The crossover made far fewer trips, 180, because two averages cross each other much less often than price crosses one. The same limits apply to these figures as to the crossover's.

If you watch for crosses anyway

  1. Read it on completed candles. Two averages that are nearly equal can cross and uncross within a day. The cross that counts is the one that stands at the close.
  2. Look at how far price has already gone. The cross is late by construction. On the day it prints, compare price with its recent high or low: that distance is what the signal has already cost or already missed.
  3. Decide in advance what a cross means for you. Sell everything, sell part, or only stop buying. A signal with no rule attached becomes a reason for whatever you already wanted to do.

Where TideTrade fits

TideTrade does not compute golden or death crosses. It runs one other kind of line, the Supertrend indicator with period 10 and multiplier 3, on 140 crypto assets, and shows each as tide in, slack water or tide out on the scanner. How that line compared with a moving average on the same data, under a test fixed in advance, is on the comparison page, with the full output and the registration. The four families of method, moving averages among them, are set side by side in the guide to trend reversals.

Questions people ask about the golden cross and the death cross

What is a golden cross in crypto?
A golden cross is the 50-day moving average of price rising above the 200-day moving average. It is read as the start of an uptrend, and it arrives after price has already risen, because both averages lag. The lengths are a convention; some traders use exponential averages or other pairs.
What is a death cross?
A death cross is the mirror image: the 50-day moving average falling below the 200-day moving average. It is read as the start of a downtrend, and it arrives after price has already fallen some distance from its high.
Is a golden cross a buy signal?
It is a statement that the 50-day average is above the 200-day, not a prediction. In one test on the 20 most-traded crypto assets, 2019 to 2026, holding from each golden cross to the next death cross returned 14.4% a year with a worst fall of 72%, against 38.5% and 85% for holding the basket throughout. That is past data from one market and one period.
How reliable is the death cross in crypto?
In the same test there were 180 round trips, each opened by a golden cross and closed by a death cross, and 72% of them lost money. The average trip lasted about 130 days. The rule was long or cash; selling short on a death cross was not tested, and no test of chance was run on these figures.
Does the golden cross work for Bitcoin?
This site has not measured Bitcoin alone. Its figures are for a basket of the 20 most-traded crypto assets, chosen again each quarter, and one asset can differ from a basket. Any claim about Bitcoin's golden crosses needs its own test with its own costs and dates.
What is the difference between a golden cross and the 200-day moving average?
The golden cross compares two averages, the 50-day and the 200-day. The 200-day rule compares price itself with one average. In the test the crossover made 180 round trips and the 200-day rule 926, 70% of them lost within ten days, because price near a single average steps over it and back.

TideTrade is decision support, not financial advice. Cryptoassets are high-risk; trend rules lose money in sideways markets and past results do not predict future prices. See the disclaimer. Every figure on this page is read from the comparison's published output.