Position sizing is the most consequential decision most investors never consciously make. Two people can buy exactly the same stocks in the same order and end a decade in completely different financial positions, purely because of how much they put in each one.
Most investing education is about selection — which company, which fund, which sector. Selection is the visible, interesting part. But your outcome is the sum of each position’s return multiplied by its weight, and the weights are the part you control with certainty.
You cannot know in advance which of your ideas is the wrong one. You can know, in advance and exactly, how much the wrong one will cost you. That is the whole discipline: you do not control whether you are right; you control the consequences of being wrong.
It follows that sizing should be decided before you buy, at the moment you are calmest and least invested in the outcome. A size chosen after a stock has run 40% is a size chosen by your emotions.
This is the workhorse. Rather than asking “how much do I want to own?”, ask two questions with real answers:
A 2% risk budget on a stock that could plausibly fall 50% gives 2 ÷ 50 = 4% of the portfolio. The same 2% budget on a speculative name that could go to zero gives 2 ÷ 100 = 2%. The method automatically shrinks positions in riskier things, which is exactly what you want and precisely the opposite of what excitement does.
| Type of holding | Plausible decline | Size at a 1% budget | Size at a 2% budget |
|---|---|---|---|
| Broad index ETF | ~35% | 2.9% | 5.7% (usually capped by policy, not by this formula) |
| Large, profitable, low debt | ~45% | 2.2% | 4.4% |
| Cyclical or leveraged | ~65% | 1.5% | 3.1% |
| High-multiple grower | ~70% | 1.4% | 2.9% |
| Speculative / pre-profit | 100% | 1.0% | 2.0% |
The risk-budget method already handles this implicitly, since more volatile holdings have larger plausible declines. But there is a cleaner version worth knowing, because it is what institutional risk desks actually do: size positions so that each one contributes equally to portfolio risk.
A stock with 60% volatility gets half the weight of one with 30%. In an equal-weighted portfolio, the most volatile names quietly dominate the portfolio’s actual risk even though they are the same size on the statement — a 5% position in something that moves twice as much is, in risk terms, a 10% position. Inverse-volatility weighting fixes that mismatch.
Its weakness: volatility is backward-looking, and a quiet stock is not necessarily a safe one. It is a useful cross-check on the risk-budget number, not a replacement for judgement about the business.
Equal weighting puts the same amount in every holding. It is simple, it needs no forecast of your own accuracy, and it is remarkably hard to beat. Its honesty is the point: it assumes you do not know in advance which idea is best, which is almost always true.
Conviction weighting puts more into higher-confidence ideas. It is defensible, and it is also where confidence and competence get confused. The uncomfortable evidence is that self-reported conviction correlates weakly with outcome — the ideas people are most certain about are frequently the ones where they have stopped looking for disconfirming information.
The Kelly criterion gives the bet size that maximises the long-run growth rate of capital, given a known edge:
With a 60% chance of a 1:1 outcome: f* = (0.6 × 1 − 0.4) ÷ 1 = 0.20 — bet 20% of capital. That number should immediately look alarming, and it should, for three reasons.
Which is why practitioners who use Kelly at all use quarter- to half-Kelly. The genuine value of the formula for a private investor is not the number it produces — it is the structural insight: optimal bet size rises with edge and falls with the chance of ruin, and there exists a size above which more confidence makes you poorer.
Formulas produce numbers; policy keeps you honest when the formula’s inputs are your own optimism. A reasonable default set for a self-directed portfolio:
| Limit | Typical cap | Reasoning |
|---|---|---|
| Single stock, at purchase | 5% | A 50% fall costs 2.5% of the portfolio — survivable and forgettable. |
| Single stock, after appreciation | 10% | Let winners run, but trim above this. See Module 9. |
| Speculative positions, in total | 5–10% | Your entire “lottery ticket” sleeve, not per name. |
| Single sector | 25% | Correlations inside a sector run 0.7–0.9 — it is closer to one position than to several. |
| Employer stock | 5% | Your salary is already fully exposed to this company. Module 6. |
These are defaults, not laws. A 20% position in a broad global index fund is fine; a 20% position in one company is a different animal entirely, because the index cannot go to zero and the company can. What matters is that your caps are written down before the moment you want to break them.
A position falls 30%. Do you buy more? This is where sizing discipline is actually tested, and the answer depends entirely on a distinction you must make honestly.
Averaging down is buying more because the price fell and you now think it is better value. Legitimate only if the thesis is intact and the fall was about price rather than about the business.
Doubling down is buying more to lower your average cost so that a smaller recovery gets you back to break-even. This is about your entry price, which the market does not know and does not care about. It is the single most reliable way to convert a manageable loss into an unmanageable one.
Educational purposes only; not financial advice. Historical figures are illustrative and past drawdowns are not a forecast of future ones. Always do your own research and consult a licensed advisor.