Does the 200-day moving average strategy work?

It lowers risk more than return. On SPY since 1993 it cut the worst fall from 55% to 23% and earned 8.9% a year against 10.7% for holding, with a Sharpe ratio statistically the same as holding.

Does the 200-day rule miss the best days?

Yes, and the worst ones too. Over the last 20 years it was in cash for all 10 of SPY’s best days and all 10 of its worst, because both come in bear markets.

Does it work on individual stocks?

It cuts the losses but costs more return. On the 500 largest US stocks it cut the share that fell by half from 34% to 10%, while the median five-year return fell from 68% to 28%.

Market factPublished Nov 2026 · SPY and FRED to Nov 6, 2026; company prices to Sep 25, 2026

The 200-day rule sat out all 10 best and all 10 worst days of the last 20 years

Since 1993, holding SPY only above its 200-day average cut the worst fall from 55% to 23% and earned 8.9% a year against 10.7%. Almost all of its edge came from two bear markets.

10 best and 10 worst days spent in cash: 20 of 20
Worst fall, rule: −23%
Worst fall, held: −55%
Return, rule: 8.9%/yr
Return, held: 10.7%/yr

What we found

Since November 1993, holding SPY only while it closed above its 200-day average, and T-bills otherwise, earned 8.9% a year against 10.7% for always holding, with a worst fall of 23% against 55%. The Sharpe ratios were 0.56 and 0.50; a year-block bootstrap puts the difference between −0.21 and +0.32, so risk-adjusted it is a draw.

It was out of the market for all 10 best and all 10 worst days of the last 20 years, and for 85% of the days SPY moved 3% or more, while out only 21% of the time. The days it sat out were twice as volatile as the days it was in (30% against 14% a year) with the same average return: the rule sidesteps volatility, not losses.

Two bear markets made its whole lead: 2000–02 and 2008–09. In the other 29 years it trailed by 4.4% a year, and it was ahead in only 5 of 34 years. Most signals were false alarms: 79% of its sales were bought back higher, and the median round trip lost 0.2%.

$10,000 in SPY since 1993, always invested against the 200-day rule, on a log scale with the rule’s time out of the market shaded; below, the share of days it was out: 21% of all days, 85% of 3% days, and all of the 10 best and 10 worst.

The full results

SPY with dividends, 1993–2026

Measure	200-day rule	Always invested
Return a year	8.9%	10.7%
$10,000 became	$168k	$283k
Volatility	12.2%	18.7%
Worst fall	−23%	−55%
Sharpe ratio	0.56	0.50
Time in the market	77%	100%
Switches a year	6.5	0

Sharpe difference and its year-block bootstrap 95% range: +0.06 (−0.21 to +0.32).

S&P 500 since 1928, prices only

Measure	Rule, cash earns nothing	Rule, cash in T-bills	Always invested
Return a year	6.6%	7.7%	6.1%
Worst fall	−54%	−54%	−86%
Sharpe ratio	0.32	0.41	0.24

The index has no dividends, so the rule is shown two ways: cash earning nothing understates it, cash in T-bills overstates it (holding loses its dividends, cash keeps its interest). T-bill data starts in 1934.

The rule against holding, a year, by decade

Cash earns nothing
Cash in T-bills

S&P 500 prices. The two bars bracket the truth: the index has no dividends.

The rule against holding, by decade

Decade	Edge a year, cash earns nothing	Edge a year, cash in T-bills
1920s	+9.5%	+9.5%
1930s	+8.6%	+8.6%
1940s	+1.3%	+1.4%
1950s	−1.8%	−1.3%
1960s	+1.2%	+2.7%
1970s	+0.6%	+3.8%
1980s	−0.4%	+2.3%
1990s	−4.0%	−3.1%
2000s	+4.6%	+5.8%
2010s	−4.5%	−4.4%
2020s	−3.8%	−3.4%

S&P 500 prices. On SPY with dividends, 2000–02 and 2008–09 made the rule’s whole lead; in the other 29 years it trailed by 4.4% a year and was ahead in 15% of all years.

How far the index had fallen when the rule sold

The whole fall
When the rule sold

How far the index had risen from its low when the rule bought back

Through the crashes

Fall	Index fall	Rule sold	Index down when sold	Rule bought back	Index up from the low
1929 crash	−86%	10.25.1929	−19%	08.10.1932	+68%
1973-74 bear	−48%	03.01.1973	−7%	01.31.1975	+22%
1987 crash	−34%	10.19.1987	−16%	06.08.1988	+18%
Dot-com bust	−49%	04.18.2000	−8%	03.25.2003	+11%
2008 crisis	−57%	11.09.2007	−6%	06.03.2009	+40%
COVID crash	−34%	03.02.2020	−13%	05.29.2020	+35%
2022 bear	−25%	01.25.2022	−8%	12.02.2022	+14%
2025 tariff drop	−19%	03.12.2025	−9%	05.14.2025	+18%

S&P 500 prices; the rule trades at the close before the date shown. In every fall it sold before most of the loss, and it bought back well above the low.

Trades on SPY

Measure	Value
Round trips	107
Won	44%
Median trip	−0.2%
Average win	+9.2%
Average loss	−1.8%
Top 5 trips’ share of the gains	72%
Sales bought back higher	79%

A sale is bought back higher when the price at re-entry beats the sale price plus the interest earned meanwhile.

Single stocks: the 500 largest each January, held 5 years

200-day rule
Held

Single stocks: the 500 largest each January 2011–2021, held 5 years

Measure	200-day rule	Held
Median 5-year return	+28%	+68%
Fell by half at some point	10%	34%
Lost half over the 5 years	1.7%	4.7%
Median worst fall	−33%	−43%
Median Sharpe ratio	0.30	0.48
Stocks that tripled: median return	+154%	+271%

5,462 stock-cohorts, 753 companies, point-in-time market caps. The rule beat holding’s Sharpe ratio for 23% of them. Survivors only.

The rule’s edge over holding, a year, by macro regime

S&P 500 since 1967
SPY since 1993

The rule by macro regime

Regime	Edge a year, S&P 500 since 1967	Edge a year, SPY since 1993	Time out of the market
Goldilocks	−3.5%	−6.9%	18%
Overheating	+0.2%	−3.2%	32%
Stagflation risk	+4.0%	+2.7%	39%
Slowdown	−1.4%	−2.8%	29%

Regime as on the /macro page, as known at the time. Edge: the rule’s average daily return minus holding’s, a year.

Why it misses the best days

The biggest up days come in bear markets, usually within weeks of the biggest down days, when the index is already well below its 200-day average. The rule is in cash then, so it misses both. Over the last 20 years the days it sat out averaged the same return as the days it held, at twice the volatility.

That is the trade it offers: a smoother ride, not a higher return. It cut SPY’s worst fall from 55% to 23% and its volatility by a third, and gave up 1.8 points of return a year to do it.

Crashes and false alarms

It got out early in every major fall: in October 1929 after 19% of an 86% fall, in 2007 after 6% of 57%, and even before Black Monday in 1987. Fast crashes cost it more on the way back: it bought back 35% above the 2020 low and 40% above the 2009 low.

Most of its signals were false alarms. 79% of its sales were bought back higher than they were sold, and 44% of its round trips made money. A few long trends paid for the rest: the top five trips made 72% of its gains.

On single stocks and by regime

On the 500 largest US stocks, the rule cut the share that fell by half from 34% to 10%, but the median five-year return from 68% to 28%, and the stocks that tripled made 154% instead of 271%. For most stocks it lowered the Sharpe ratio.

By the /macro page’s regime, the rule added 4.0% a year in Stagflation-risk months and cost 3.5% a year in Goldilocks months, since 1967. It pays when inflation and weak growth come together, and charges rent the rest of the time.

How it was measured

SPY with dividends from 1993. In while the close is above its 200-day simple average; otherwise cash at the 3-month T-bill rate (FRED TB3MS).
A signal on one day’s close is traded at the next day’s close, as in the QuantLabs backtester, and every switch costs 0.05%.
The S&P 500 index from 1928 for the long view. It has no dividends, so the rule is shown with cash earning nothing and with cash in T-bills, which bracket the truth.
Single stocks: the 500 largest US companies by point-in-time market cap each January from 2011 to 2021, each followed for five years.

Caveats

The single-stock data holds only companies still listed today, so stocks the rule would have sold on their way to bankruptcy are missing; its protection there is understated.
No taxes. In a taxable account, 6.5 switches a year would realise gains that holding defers.
Since 1928 its edge leans on the 1930s: it sold in October 1929 and missed most of the 86% fall. It trailed holding in the 1950s, 1990s, 2010s and 2020s.

Questions

Does the 200-day moving average strategy work?: It lowers risk more than return. On SPY since 1993 it cut the worst fall from 55% to 23% and earned 8.9% a year against 10.7% for holding, with a Sharpe ratio statistically the same as holding.
Does the 200-day rule miss the best days?: Yes, and the worst ones too. Over the last 20 years it was in cash for all 10 of SPY’s best days and all 10 of its worst, because both come in bear markets.
Does it work on individual stocks?: It cuts the losses but costs more return. On the 500 largest US stocks it cut the share that fell by half from 34% to 10%, while the median five-year return fell from 68% to 28%.

More research

All research →

Historical analysis. Past performance does not guarantee future results. Not investment advice. Published Nov 2026. Data: SPY and FRED to Nov 6, 2026; company prices to Sep 25, 2026.