Technical analysis8 min read

Does the Golden Cross Work? 100 S&P 500 Crosses Since 1928, Tested

A tiny edge that luck matches, and a rule that tied holding over a century with half the worst fall.

By Pavel Penev, MScFounder, TradeWize · 10+ years trading the markets

The short answer

Not as a forecast. A year after a golden cross, the S&P 500 was higher 74% of the time. After any day it was 70%, and random days match that edge 31% of the time. A death cross did the same as any day. Trading the crosses roughly tied just holding over 98 years, with about half the worst fall. Since 2010 it made less than half as much.

A golden cross is when the 50-day moving average climbs above the 200-day. A death cross is the opposite: the 50-day drops below the 200-day.

They're the most famous signals in charting, and the news covers every one. Our moving-average guide explains how they work on a made-up chart. This post tests them on the real thing: every cross the S&P 500 has printed since both averages began in October 1928, up to 19 November 2026.

100
Crosses since 1928: 50 golden and 50 death.
74% vs 70%
How often the S&P 500 was higher a year later: after a golden cross, and after any day.
9.9% vs 10.0%
A year, trading the crosses vs holding, since 1928, dividends in.

Finding 1: every cross comes late

The S&P 500 has had 100 crosses since 1928, 50 of each. They alternate: a golden cross, then a death cross, then a golden cross again.

The S&P 500 and all 100 crosses
ALL 100 CROSSES SINCE 1928The S&P 500 at each month-end on a log scale, with a dot on every golden and death cross.101001k10k19301950197019902010golden cross (50)death cross (50)S&P 500 price, Yahoo Finance, to 19 Nov 2026. Dots sit on the day's close.

Month-end closes on a log scale. Green dots are golden crosses, red dots death crosses.

Each cross is late by design. A 200-day average is the last ten months of prices, so it turns slowly. By the time the 50-day crosses it, the move is well under way.

The median golden cross came 64 trading days after the low, about three months, with the index already up 15%. The median death cross came 57 trading days after the top, with the index already down 9%.

The first one shows how late it can be. The 50-day stayed above the 200-day right through the crash of October 1929. The first death cross came on 22 November 1929, with the S&P 500 already 32% below its September top.

Crosses don't flip back and forth as often as you might think. Only eleven of the 99 finished crosses were undone by the opposite cross within 40 trading days. The median cross lasted 180 trading days, about nine months.

Finding 2: a death cross often comes near the bottom

The name suggests a death cross means more pain ahead. Sometimes it did. More often, most of the fall was already over.

Twenty of the 50 finished death crosses came with less than 5% left to fall before the next golden cross. Selling on those meant selling close to the low. Another 21 had 5% to 20% left.

How much further the S&P 500 fell after each death cross
HOW MUCH FURTHER IT FELL AFTER A DEATH CROSSAll 50 finished death crosses, by the fall from the cross to the low before the next golden cross.under 5%20 crosses5–10%13 crosses10–20%8 crosses20% or more9 crossesS&P 500 daily closes, price only. Red: 1929, 1937, 1973, 2000 and 2007 are among them.

From the close on the day of the cross to the lowest close before the next golden cross.

Nine came before another fall of 20% or more: 1929, 1937, 1940, 1962, 1969, 1973, 1981, 2000 and 2007. Those are the ones people remember. The 1929 death cross was followed by a further 80% fall, to the low of 1 June 1932. The one in December 2007 came before another 54%.

Golden crosses split too. Eight of the 49 finished ones led to a run of 50% or more before the next death cross. The one in September 1994 ran 150%, to July 1998. But eleven came within 5% of the top. The April 2019 one was still in place when COVID hit, and the index was 10% lower a year later.

Finding 3: a year out, a golden cross was close to any day

Here's the test that matters. If a golden cross is good news, the S&P 500 should do better after one than after a random day.

It did, a little. A year after a golden cross, the index was higher 37 times out of 50. That's 74%, against 70% for every day since 1928. The median gain was 12.7%, against 9.4%.

That edge is small enough to be luck. We picked 50 random days 20,000 times. 31% of those picks had at least 37 days higher a year later. 11% had a median at least as high. With 50 crosses, one cross moves the up rate by two points.

Death crosses were higher a year later 34 times out of 50, or 68%. That's the same as any day. The short run was worse: three months out, the index was higher after only 54% of death crosses, against 64% of all days.

What came after each kind of cross
Higher 3 months laterHigher 6 months laterHigher a year laterMedian, a year
Golden cross70%72%74%+12.7%
Death cross54%54%68%+8.6%
Any day64%67%70%+9.4%

S&P 500 price change from the close on the day, 1928 to 2026. Since 1950 alone, golden crosses were higher a year later 82% of the time against 75% for any day.

Learn it by doing

Reading about it is one thing — it clicks when you do it. Learn it hands-on with free, interactive lessons on TradeWize.

Try the free lesson →

Finding 4: trading the crosses tied holding, with smaller falls

The classic way to trade the crosses is simple. Hold the S&P 500 while the 50-day is above the 200-day. After a death cross, sell and sit in Treasury bills until the next golden cross.

We tested it from October 1928, with dividends counted for both sides. $1 held the whole time grew to $11,094. $1 that followed the rule grew to $10,590. That's 9.94% a year against 9.99%, close to a tie.

Holding vs the cross rule since 1928
THE CROSS RULE AGAINST JUST HOLDING$1 from Oct 1928, dividends in, log scale. The rule sits in T-bills after a death cross.$1$10$100$1k$10k19301950197019902010buy and hold: $11,094cross rule: $10,590Shiller dividends to 1987, S&P 500 Total Return index after. No fees or taxes.

Growth of $1 on a log scale, dividends in. The rule holds T-bills after a death cross and buys back at the next golden cross. No fees or taxes.

The rule's real gain was smaller falls. Its worst was 46%, against 84% for holding through the Great Depression. It sat out 32% of days, and the T-bill interest it earned there did a lot of the work. Without it, the rule's $1 grew to only $3,798.

The start date changes the answer a lot. The rule trailed holding from four of the five later starts we tried. From 2010 it made 4.0x against 8.6x, 46% of what holding made, with the same worst fall.

The cross rule against holding, from six start dates
StartCross ruleBuy and holdRule's worst fallHold's worst fall
October 192810,590x (9.9%/yr)11,094x (10.0%/yr)−46%−84%
January 19503,550x (11.2%/yr)4,575x (11.6%/yr)−34%−55%
January 1970364x (10.9%/yr)369x (10.9%/yr)−34%−55%
January 199029.6x (9.6%/yr)41.7x (10.7%/yr)−34%−55%
January 20008.3x (8.2%/yr)8.1x (8.1%/yr)−34%−55%
January 20104.0x (8.6%/yr)8.6x (13.7%/yr)−34%−34%

Growth of $1 to 19 November 2026, dividends in, T-bills when out. Fees and taxes would cost the rule more than holding: it switched 100 times.

2020 shows why the rule can't dodge a fast crash. The S&P 500 peaked on 19 February 2020 and hit bottom on 23 March 2020. The rule held the whole way down. The death cross came on 30 March 2020, a week after the low, at 2,627. The rule bought back on 9 July 2020, at 3,152.

That's why its worst fall since 1950 is the 2020 crash, 34%, the same as holding. It helps in slow bear markets like 2000-02 and 2008, and does little in fast ones.

What people say

The crosses call the turns.

  • A golden cross means a bull market is starting.
  • A death cross means get out now.
  • Trade the crosses and you'll beat holding.

What the data says

Late signals, small edge.

  • A year out, 74% of golden crosses were higher. Any day: 70%.
  • Twenty of 50 death crosses had less than 5% left to fall.
  • Since 1928, the rule made 9.9% a year. Holding made 10.0%.

Finding 5: where the S&P 500 sits now

The last death cross came on 14 April 2025, during the tariff sell-off, at 5,406. It was one of the near-the-bottom ones. The low came a week later, on 21 April 2025, just 5% lower.

The golden cross followed on 1 July 2025, at 6,198. As usual, it was late: the index was already 20% above the April low. A year later it was up 21%.

The S&P 500 since 2018, with both averages
THE LAST 8 CROSSESThe S&P 500 since 2018, with its 50-day and 200-day moving averages.2,0004,0006,0008,0002018202020222024202650-day200-daygolden crossdeath crossS&P 500 closes, every fifth trading day, to 19 Nov 2026.

Close, 50-day and 200-day averages. The 2020 death cross came a week after the low.

On 19 November 2026, the S&P 500 closed at 7,176. That's 8.2% below its high of 6 October 2026, and below its 200-day average (7,348) since 17 November 2026. The 50-day is still 4.0% above the 200-day, so there's no death cross yet.

If one comes, the record above is the best guide we have. It might mark the start of a real bear market. More often, most of the fall was already done.

How to read it

  • Treat a cross as a summary of what already happened, not a forecast.
  • Don't panic-sell on a death cross. In 40% of cases the fall was nearly over.
  • Don't buy because of a golden cross either. A year later it did about as well as a random day.
  • If you use the rule, it's for smaller falls, not bigger returns. It needs a tax-free account, cheap trades and the patience to sit in T-bills while the market rises.
  • If you invest a fixed amount every month, you don't need to watch the crosses at all.

What is a golden cross in stocks?

It's when a stock or index's 50-day moving average rises above its 200-day moving average. The S&P 500's latest was on 1 July 2025, at 6,198.

Is a golden cross bullish?

A little, and not reliably. A year after a golden cross the S&P 500 was higher 74% of the time, with a median gain of 12.7%. After any day it was higher 70% of the time, with a median gain of 9.4%. Random days match that edge 31% of the time.

Does a death cross mean a crash is coming?

Usually not. Twenty of 50 S&P 500 death crosses came with less than 5% left to fall. Nine came before another 20% or more, including 1929, 2000 and 2007. A year later the index was higher after 68% of them, the same as any day.

Does trading golden and death crosses beat buy and hold?

Not on returns. Since 1928, with dividends in and T-bills when out, the rule made 9.9% a year against 10.0% for holding. Its worst fall was 46% against 84%. From 2010 it made 46% of what holding made. This isn't advice.

Method, and what this can't tell you

S&P 500 daily closes from Yahoo Finance, 1927-12-30 to 2026-11-19. The 50-day and 200-day lines are plain averages of the last 50 and 200 closes, so both exist from 1928-10-16. A golden cross is the first close with the 50-day above the 200-day; a death cross is the first close with it below. "Late" is measured from the lowest close (or highest, for a death cross) since the previous cross. Forward changes are 63, 126 and 252 trading days, price only, compared with every trading day since the lines began. The rule test holds the S&P 500 from the day after a close with the 50-day above the 200-day and 3-month Treasury bills otherwise, with dividends on both sides: Robert Shiller's monthly dividends to 1987, the S&P 500 Total Return index from 1988.

  • Before 1957 the S&P index had 90 stocks, not 500. Yahoo's series joins the two.
  • Closes only. A site using an exponential average, or a different data feed, can date a cross a few days differently.
  • Fifty crosses on each side is still a small sample. One cross moves a percentage by two points.
  • The rule test has no trading fees, no taxes and no slippage. In a taxable account, the 100 switches would cost it more than holding.
  • Dividends before 1988 are spread evenly through each month. T-bill rates before 1934 use January 1934's rate.
  • None of this is a forecast. The next cross doesn't have to behave like any of these.

Learn to read a chart

Our technical analysis track teaches trends, moving averages and patterns step by step, with charts you practise on.

Written by

Pavel Penev, MSc

MSc Investment & Finance, Queen Mary University of London · 10+ years trading the markets

Pavel founded TradeWize after years of trading and an MSc in Investment & Finance from Queen Mary University of London. He writes these guides to teach the decisions, not just the theory.

More about TradeWize →

Terms in this article

Keep reading