Guide
Planning an account, not just picking stocks
A signal is about one stock. What happens to your money depends on the mix: how many names, how much in each, how much spread across everything, and whether you trim what grows. Here is what those choices did since 2006, from 2,000 random portfolios at each setting, built from the 221 covered stocks that have traded the whole time.
1. How many stocks
Each row is 2,000 portfolios of that many covered stocks picked at random, equal amounts in each, from 2006 to October 2026 (21 years). The last row is all 221 together.
| Holding | Typical yearly return | Bad luck to good luck (10th–90th) | Typical worst fall | Fell 50%+ at some point | Beat all 221 together |
|---|---|---|---|---|---|
| 1 stock | 10.4% | 4.5% to 18.3% | −68% | 82% | 13% |
| 2 stocks | 14.3% | 8.8% to 20.6% | −60% | 73% | 24% |
| 3 stocks | 15.2% | 10.6% to 21.0% | −57% | 69% | 28% |
| 5 stocks | 16.1% | 12.1% to 20.8% | −53% | 59% | 33% |
| 10 stocks | 17.0% | 13.9% to 20.3% | −49% | 45% | 38% |
| 20 stocks | 17.3% | 15.2% to 19.6% | −47% | 30% | 41% |
| 30 stocks | 17.5% | 15.7% to 19.4% | −47% | 23% | 43% |
| All 221, equal amounts | 17.7% | – | −45% | No | – |
- One stock is a lottery ticket. The typical single stock returned 10.4% a year, against 17.7% for all 221 together. One in ten did 4.5% or worse and one in ten 18.3% or better, and 82% fell by half or more at some point. Only 13% beat the whole list: a few huge winners carry the average, and most stocks aren't them.
- The spread closes fast. By 10 stocks the bad-luck case was 13.9% a year; by 30 it was 15.7%. The typical worst fall went from −68% with one stock to −47% with 30, close to the −45% of all 221.
- Most of the protection comes from the first 10 to 20 names. Past that, more stocks change the result little, and an index fund does the same job in one holding.
These are today's survivors: stocks that went bankrupt or were bought out aren't in the list, which flatters every row, and single stocks most. That's also why all 221 together returned 17.7% a year against 11.0% for the S&P 500 (SPY), which includes the companies that failed. Each stock is in its own currency.
2. A core with a slice of picks
Here the picks are three covered stocks chosen at random, the way a short "top picks" list might look, and the core is all 221 stocks in equal amounts, so picks and core come from the same survivors. Each mix was reset to its split every week.
| Mix | Typical yearly return | Bad luck (10th) | Typical worst fall | 2008–09 financial crisis | 2020 COVID crash | 2022 rate shock |
|---|---|---|---|---|---|---|
| All picks | 15.2% | 10.5% | −56% | −44% | −34% | −17% |
| 75% picks, 25% the rest | 16.2% | 12.6% | −52% | −43% | −34% | −16% |
| 50 / 50 | 17.0% | 14.5% | −48% | −42% | −34% | −16% |
| 25% picks, 75% the rest | 17.5% | 16.2% | −46% | −42% | −35% | −16% |
| 10% picks, 90% the rest | 17.6% | 17.1% | −45% | −41% | −35% | −16% |
| All 221, equal amounts | 17.7% | – | −45% | −41% | −35% | −16% |
- Picks didn't add return on average. The typical three-stock slice returned 15.2% a year against 17.7% for the whole list. What picks add is range: the chance of a big winner, and of a big loser.
- The core lifts the floor. Going from all picks to 25% picks took the typical worst fall from −56% to −46% and lifted the bad-luck case from 10.5% to 16.2% a year.
- In a broad crash, everything falls. In 2008–09 the typical mix fell 40% or more whatever the split; the core shrank the bad-luck cases more than the typical one.
- Left alone, a split drifts. A 50/50 mix never reset ended with the picks typically at 39% of the account; when a pick takes off the drift runs the other way, and a balanced plan quietly becomes a concentrated one.
3. One stock that's most of the account
2,000 accounts with 60% in one covered stock and 10% in each of four more, run four ways:
| Approach | Typical yearly return | Bad luck (10th) | Typical worst fall | Worst fall, bad luck (10th) | Fell 50%+ |
|---|---|---|---|---|---|
| 60% in one stock, never trimmed | 14.1% | 10.2% | −56% | −74% | 68% |
| 60% in one stock, trimmed back to 60% | 14.6% | 10.2% | −56% | −79% | 66% |
| That stock capped at 25%, the rest spread across all | 16.8% | 14.1% | −48% | −60% | 43% |
| Capped at 10%, the rest spread across all | 17.3% | 15.2% | −47% | −55% | 30% |
- A cap changed the bad cases most. Capping the big stock at 25% and spreading the rest took the share of accounts that fell by half or more from 68% to 43%, and the bad-luck worst fall from −74% to −60%.
- Letting it ride is how the occasional huge winner happens, and also how most of the 50% falls happened.
- Trimming back to 60% didn't help: topping a falling stock back up to 60% made the worst cases worse (−79%). The size of the position mattered more than the trimming.
4. What the professionals are held to
- In Canada, a mutual fund generally can't buy a security if that would put more than 10% of its net assets in one issuer (National Instrument 81-102, section 2.1; as the OSC describes it).
- In the U.S., a fund that calls itself diversified must have at least 75% of its assets in holdings where no single issuer is more than 5% of the fund (Investment Company Act of 1940, section 5(b)(1)).
- Those rules are for funds holding other people's money. They're a reference point, not a rule for anyone's own account.
5. The same bet twice
Two stocks in the same business tend to fall together. Over the last three years, covered chip stocks moved together far more than two covered stocks picked at random (the numbers are in One stock too many), so a list with five semiconductor names behaves more like one or two positions than five. The account check on the watchlist shows how much of your list sits in each sector.
What this can and can't tell you
It shows what each choice did in the past, on stocks that survived. It can't say what mix is right for you: that depends on when you need the money, what else you own, your taxes and how you'd handle a 40% fall, which is what an advisor is for (ways to invest, and how to check an advisor, by country). This analysis does not constitute trading advice. Please meet with an advisor or independently review sources before making any decision.
Study run 2026-10-13: weekly closes adjusted for splits and dividends, no trading costs or taxes, 2,000 random draws per setting.