Allocation Strategy
← Portfolio · conviction scoring lives on the Four Key Squares pageA Bollinger band drapes two standard deviations around a stock's recent average price; "%B" says where today's price sits inside that band — 0 at the bottom, 1 at the top. We measured this on three clocks at once (daily, weekly, monthly) for every liquid Japanese stock, every day, for five years, then checked each position against the price 12 months later. All combinations, two matrices:
The two matrices differ only in the monthly trend — and that difference is the strategy. The best cell in the book is a sharp daily dip inside a healthy long trend (up to 73 of 100 higher a year later). The worst is a bounce above the daily band while the long trend is broken (41 of 100). The portfolio's three ranges are these cells translated into yen for each holding.
The four situations, in plain numbers
| Out of 100 stocks that were… | Higher 12 months later | Typical move |
|---|---|---|
| A dip on the daily chart while the monthly trend is healthy | 61 of 100 | +7.4% |
| Monthly trend healthy, nothing stretched, no dip | 62 of 100 | +7.7% |
| Price above its weekly band | 59 of 100 | +5.8% |
| Monthly trend broken | 53 of 100 | +1.9% |
And any stock, any position, held for…
| Holding period | Higher at the end | Typical move |
|---|---|---|
| 6 months | 58 of 100 | +3.5% |
| 12 months | 60 of 100 | +6.6% |
| 18 months | 62 of 100 | +9.1% |
| 24 months | 65 of 100 | +14.7% |
From study to weights
The worked example, step by step. Take 7378 ASIRO. Every name starts equal. First factor: its Four Key Squares. ASIRO's squares are still contested — not won, not lost — so this factor is 1.0. It earns no bonus and takes no penalty. Second factor: valuation. Among the holdings, ASIRO's price is the lowest relative to the profit its capital produces. The cheapest gets the biggest reward, so this factor is 1.15. Third factor: the tape. ASIRO's current position in its bands belongs to a group that, historically, rose more often than average over the following 12 months. That earns 1.11. Multiply the three: 1.0 × 1.15 × 1.11 ≈ 1.28. Do the same for every holding, scale them so the whole portfolio adds up to 100%, and ASIRO's share comes out at 6.80%.
The fine print. The study behind the matrices covers 2.4 million daily observations from 2021 to 2026. We kept only stocks that trade at least ¥10 million a day — stocks an investor could actually buy. Returns are measured on prices alone; dividends are left out, and they would only have helped. The same stock counts once per day, so observations overlap and the sample is large but not independent. The weekly and monthly readings use only finished bars — when we ask "what happened over the next 12 months," everything we knew at the start was truly knowable then. And the honest limit: these numbers describe the past five years. They do not promise the next five.
Who makes the team. As more companies get profiled, the portfolio does not grow with them. Every published company is scored with the same three factors. The twenty highest scores hold a place; the rest wait outside. One exception overrides everything: a company whose valuation re-rates out of the screen leaves immediately, however well it scores.
What the tape never decides. The bands never answer whether a company is worth owning. That question is answered on the Four Key Squares page, from filings. The tape answers only two smaller questions: how quickly to build the position, and when to stand still.