STOCKS · 30 YEARS · PUBLISHED 28 JUL 2026

Thirty years of data on the famous timing rules

Four rules that get repeated endlessly in trading forums, books, and YouTube thumbnails. Thirty years of dividend-adjusted S&P 500 data. Every one of them lost to doing nothing.

The claims we tested

Each of these is stated as settled wisdom somewhere on the internet right now: “Never hold below the 200-day.” “The golden cross is the most reliable signal in markets.” “Sell in May and go away.” “Buy when RSI is oversold.” They are all testable, and almost nobody publishes the test.

Method

SPY daily closes over 29.2 years (7,347 trading days), adjusted for dividends so the buy-and-hold benchmark is honest — comparing a timing strategy against a price-only benchmark quietly hands the strategy several points a year. Each rule is fully invested or fully in cash, charged 0.05% per switch. Every signal is computed from data available at the previous close, then the next day's return is applied — and every strategy, including the benchmark, starts from the same bar.

That last sentence is the whole ballgame. Our first run let each rule read the same day's closing price and then collect that day's move — and it reported the 200-day rule earning 18.6% a year. Fixing that one index dropped it to 7.18%. We wrote up exactly how that happened →

Results

Strategy	Total	CAGR	Worst DD	In market	Switches
Buy & hold S&P 500 (SPY)	1,364%	9.62%	−55.2%	100%	0
Golden cross (50/200)	1,008%	8.58%	−33.7%	75%	29
200-day MA timing	692%	7.34%	−27.7%	76%	187
“Sell in May” (Nov–Apr only)	524%	6.47%	−36.7%	49%	58
Buy the dip (RSI<30, hold 20d)	273%	4.61%	−42.1%	26%	176

Source: Yahoo Finance adjusted closes. Costs 0.05% per switch. No leverage, no shorting.

Correction, 29 July 2026. The figures in this table were updated. The original version measured each timing rule from its own warm-up bar while the buy-and-hold benchmark started ~200 trading days earlier — so the benchmark was credited with extra compounding the rules never had, overstating it by about 0.8 points a year. Every strategy is now measured from the same starting bar. The conclusion did not change: no timing rule beat holding, and the 200-day rule still roughly halves the drawdown. We found this while building the public tester, and correcting it publicly is the point of the whole exercise.

What the numbers actually say

Buy-and-hold won on return, by a lot. Twenty-nine years of doing nothing returned 9.62% a year. The best timing rule managed 8.58%. Over three decades that gap compounds into roughly half your final wealth.

But the 200-day rule did do something real: it cut the worst drawdown from −55.2% to −27.7%. It didn't make more money — it made the ride survivable. Whether that trade is worth 2.3 percentage points a year depends entirely on whether you would have actually held through a 55% loss. Most people find out that they wouldn't, at the worst possible moment.

Buying the dip was the worst of all — 4.61% a year, and it still ate a −42% drawdown, because “oversold” keeps getting more oversold in a real bear market.

The pattern, stated plainly: these rules are not return generators. They are insurance policies, and the premium is paid in performance. That is a legitimate thing to buy — but it is not what they are usually sold as.

Why we ran this

We ran the same tests on our own crypto trading. When we pointed this engine at seven crypto strategies with real money, all seven lost — and we published that too. The honest finding across both markets is identical: timing rules reduce pain, not increase profit.

Get the next autopsy

We test one strategy people swear by and publish whatever the data says — including when it says the strategy doesn't work. Free, no pitch.

Vorrik Research publishes tests of publicly discussed trading strategies using public market data. Where a test was also run with our own money, the article says so. It is factual reporting of what happened in historical and past live tests — not investment advice, and not a recommendation to buy or sell anything. Past results never predict future results.