PlainQuant

Famous stock picking rules, honestly tested

Does the 200-day moving average rule work?

Enter when a stock closes above its 200-day average after closing at or below it; exit when it closes back below, or 50% below the entry as a safety stop.

US 2019–2026: raw backtest 23.2% a year, honest estimate 22.5% a year. Hong Kong 2019–2026: raw backtest 0.0% a year, honest estimate -1.2% a year.

MarketRaw backtest a yearHonest estimate a yearTradesThe market, same years
US2019–2026 23.2%22.5%388—
Hong Kong2019–2026 0.0%-1.2%6231.1%

Why is the honest estimate different from the raw backtest?

US: Trading costs took 0.3% a year: 23.5% before costs, 23.2% after. We cut the test into 3 periods; the middle one returned 22.5% a year, below the whole run, so the honest estimate starts from it. Stocks that were later delisted are in our prices for this market, so nothing more is taken off.

Hong Kong: Trading costs took 3.3% a year: 3.3% before costs, 0.0% after. We cut the test into 3 periods; the middle one returned -1.2% a year, below the whole run, so the honest estimate starts from it. Our prices for this market miss most stocks that were later delisted; a loss is not cut, since that would make it look smaller.

A rule tried once, with settings fixed before the test, takes no discount for luck.

Did it work in every period?

US: It made money in 3 of 3 periods: 2019–2022 22.5% · 2022–2024 10.6% · 2024–2026 38.2%

Hong Kong: It made money in 1 of 3 periods: 2019–2022 24.3% · 2022–2024 -18.6% · 2024–2026 -1.2%

How often did a trade make money?

US: 20% of 388 finished trades made money; the middle trade returned -2.3% and was held 7 days. Without its 10 best trades it would have lost money.

Hong Kong: 16% of 623 finished trades made money; the middle trade returned -3.1% and was held 6 days.

How is this different from the golden cross?

The golden cross waits for the 50-day average to cross the 200-day average; this rule acts on the price itself crossing the 200-day average, so it trades sooner and more often.

Where does the rule come from?

Comparing a price with its 200-day (about 10-month) average is a long-standing trend filter, often linked to Mebane Faber's paper "A Quantitative Approach to Tactical Asset Allocation" (2007), which applies a moving average to whole markets. This page tests a daily version on single stocks.

How to read this

Raw backtest: after costs. Honest estimate: the lower of the whole run and its middle period, then cut where our prices miss delisted stocks (Hong Kong); a rule tried once takes no discount for luck. A free account tests the last 5 years, so its numbers can differ.

The market (Hong Kong): the Tracker Fund of Hong Kong 2800, dividends included, over the same dates.

How it works →

Test it yourself: US Test it yourself: Hong Kong