Do Popular Stock Strategies Work? A 10-Year Test on Japanese Stocks

We tested 15 well-known stock-picking strategies on every Tokyo Stock Exchange stock (including delisted ones) over about 9.2 years (2017-07-03 to 2026-09-30). After tax and trading costs, did they beat simply holding a global all-world index fund or TOPIX (the Japanese market)?

POINT

Strategies that beat the all-world index: 0 of 15

Over the same period, the all-world index returned 14.2% a year after tax, and TOPIX 11.2%.

Strategy chart

Read from the top. Tap a card to open that strategy's results. The number on each card is the annual return after tax and costs; the label on the right is the verdict against the all-world index.

Can this strategy really beat the all-world index?

↓

Stage A: Buy on price patterns (technical)

1Golden cross (25-day × 75-day)4.2%Lost 2Short golden cross (5-day × 25-day)6.8%Lost 3Buy when RSI falls below 304.4%Lost 4Buy at the lower Bollinger Band2.5%Lost 5Buy at 52-week highs6.7%Lost 620-day breakout (Turtle style)3.9%Lost 7Buy the crash, sell the rebound−10.8%Lost

↓

Stage B: Pick by financial metrics (screening)

8High dividend yield7.5%Lost 9Low price-to-book (value)12.1%Mixed 10Low P/E (value)13.5%Mixed 11High ROE (quality)−1.4%Lost 12Small caps−10.0%Lost 13Low volatility4.0%Lost 14Momentum (12-month winners)−18.7%Lost

↓

Stage C: Buy on news (event)

15Buy upward earnings revisions0.4%Lost

↓

Result: 0 of 15 strategies beat the all-world index

↓

Stage D: How to read this test

?Read the method and limitations

All strategies at a glance

Strategy	Annual return (after tax)	vs. All-World	vs. TOPIX
1 Golden cross (25-day × 75-day)	4.2%	Lost−10.0pt	Lost−6.9pt
2 Short golden cross (5-day × 25-day)	6.8%	Lost−7.4pt	Lost−4.4pt
3 Buy when RSI falls below 30	4.4%	Lost−9.8pt	Lost−6.8pt
4 Buy at the lower Bollinger Band	2.5%	Lost−11.7pt	Lost−8.7pt
5 Buy at 52-week highs	6.7%	Lost−7.5pt	Lost−4.4pt
6 20-day breakout (Turtle style)	3.9%	Lost−10.3pt	Lost−7.3pt
7 Buy the crash, sell the rebound	−10.8%	Lost−25.1pt	Lost−22.0pt
8 High dividend yield	7.5%	Lost−6.7pt	Lost−3.6pt
9 Low price-to-book (value)	12.1%	Mixed−2.1pt	Won+1.0pt
10 Low P/E (value)	13.5%	Mixed−0.7pt	Won+2.3pt
11 High ROE (quality)	−1.4%	Lost−15.6pt	Lost−12.6pt
12 Small caps	−10.0%	Lost−24.3pt	Lost−21.2pt
13 Low volatility	4.0%	Lost−10.3pt	Lost−7.2pt
14 Momentum (12-month winners)	−18.7%	Lost−32.9pt	Lost−29.9pt
15 Buy upward earnings revisions	0.4%	Lost−13.8pt	Mixed−10.7pt

"Won" means a higher annual return and winning in at least 60% of 75 three-year runs started in different months. "Mixed" means only one of these, or winning at least 40% of the runs. See Method.

About this test

The strategies, exit rules and verdict rules were fixed and recorded before any results were tallied. The numbers were computed once with data as of October 2026 and will not be updated.

Even a strategy marked "Won" may simply have been lucky in this one period. Our confidence in any win is low.

Don't believe it? Ask an AI

Copy the text below and ask ChatGPT, Gemini or Claude. Ask for counterarguments too.

According to https://toushi-kensho.pages.dev/en/, only 0 of 15 popular stock strategies (golden cross, high dividend, low P/B, momentum, etc.) beat a global all-world index fund after tax and costs over ~9.2 years of Japanese data. Is this consistent with other research and public data (e.g. S&P's SPIVA reports)? What are the strongest counterarguments?

AI answers can be wrong too, and an AI cannot rerun our backtest; use it to check consistency with other research.