At a glance
These numbers matter most. Each is explained quickly below.
- 30%Minimum contribution, of gross income. More is allowed.
- Top 15Names bought each month, brought toward equal position size.
- 10%Once a position reaches 10%, I stop adding to it. Above that level, it's trimmed back.
- 15% and 25%Caps on any one industry and any one sector. Countries and emerging markets have caps of their own, below.
- 50%Positions are sold when they fall out of the top half of my ranked list, whatever their size, unless the stock is among the past year's strongest risers.
- F score ≤ 4No F score, or one of 4 or less, and I don't buy: that low a score points to a business going downhill.
- 1 yearPositions are held for more than a year before any gain is sold.
- MonthlyI only buy or sell on the first trading day of the month. The one exception is taking money out for a medical emergency.
- 0Trade alerts, ever. There's no signals group.
How stocks are ranked
US-listed companies above the smallest fifth of NYSE companies by market value (about $1.2 billion), trading at least $3 million a day, excluding financial companies, REITs, partnerships that send a K-1, and royalty trusts, among other requirements in the full rules. Each is scored on three characteristics from the Fama-French five-factor model, and the three are averaged with equal weight. A company missing any of the three isn't ranked at all.
| Characteristic | Measured as | Ranked against |
|---|---|---|
| Value | Book value ÷ the larger of today's and last year's market value (higher is better) | Its own industry |
| Profitability | (Operating income + depreciation and amortization − interest) ÷ total assets (higher is better) | Every eligible company |
| Investment | One-year growth in total assets (lower is better) | Every eligible company |
Why value only competes within an industry. Published research finds most of book-to-market's predictive power comes from differences between companies in the same industry. Between industries, it mostly reflects different business models.
For why each score is there, the F score's nine tests, and what the screen drags in by accident, see The Screen.
Buying and selling
Buying, every month
- Take the top 15 names on the ranked list that the trap guard doesn't block.
- Skip any that has agreed to be acquired, is in bankruptcy, has withdrawn its financial statements, says it's a PFIC, or can't be bought that day. Skip, too, any at 10% of the portfolio, any in an industry at 15% or a sector at 25%, and any in a country at its cap or in an emerging market when those together are at theirs. The next name down takes its place.
- Split the contribution to bring those fifteen toward equal size. If a cap stops money, it goes to the next names down the list, so nothing sits in cash.
- If a company drops out of the top 15, I keep holding it. It gets money only when a cap sends money down the list.
- If the US market's two-year return (VTI's) goes negative, purchases switch to a ranking that values stocks at today's price.
Selling: six triggers only
- Falls below the top half of the ranking while not among the past year's strongest risers, or goes two monthly dates in a row without a rank.
- Grows past 10%: trimmed back.
- An industry past 15%, a sector past 25%, a country past its cap, or emerging markets together past theirs: trimmed back.
- A position under 0.5%, outside the top 15 and holding only lots owned for over a year: closed.
- Files for bankruptcy: I sell the whole position at the next monthly date, gain or loss.
- Admits to being a PFIC in its annual report: I sell the whole position at the next monthly date.
No gain is sold within a year of buying, for the tax treatment. The exceptions are a holding past one and a half times any cap, which is trimmed regardless, a holding in bankruptcy or that's a PFIC, and a permitted withdrawal. Leaving the top 15 is never a reason to sell. A holding bought out for cash is paid out whenever the deal closes, and any choice a holding offers gets whatever happens if I do nothing.
Country caps: 62% for the US, its share of world stock markets, and 10% for any other country. Together, emerging and frontier markets top out at 10%, their share. The two shares are FTSE's figures for September 2026, and they're reset each January.
Money only comes out for two things: once toward a home, and for a medical emergency.
Changing these rules
If I could change a rule every time it got in my way, I'd never have to count a breach. The rule just moves to allow it. So changes are slow on purpose.
- January only
Once a year, on schedule, whatever the results.
- 30 days' notice
Published before it takes effect, with the reasoning.
Fixing a factual error, or a rule that's become impossible to follow, is allowed at any time, and so is publishing the options account's rules. Any other change outside the window counts as a breach. It's recorded and never removed. Some things can never change at all: the index starting from 100, time-weighted; a second account, if one is added, reported separately; the Breach Count, never reset; every version, kept; and no trade alerts, ever.
Version history
| Version | In force | What changed | Full text |
|---|---|---|---|
| 1.0 | From November 1, 2026 | Version 1.0, written and dated before my first trade. | Read |
I never take a version down, and each one stays up in full. The history is also in a public repository, so the dates can be checked independently.
Download this version