How Much of Each Stock Should I Own? 4 Weighting Methods
Equal weight, inverse volatility, mean-variance optimization and risk budgeting compared on one sample portfolio, with the backtest record of each. Education, not a recommendation.
Position sizing decides how big each holding can get. Here are two ways to set a size from data, by volatility and by crash loss, plus the caps funds and investors commonly use.
What is your own portfolio exposed to? Check it with Exent →
Position sizing is deciding the maximum share of your portfolio any one holding may take, based on how much that holding could cost you. It is separate from picking what to own. Two investors can own the same ten stocks and have very different risk, because one put 30% in the most volatile name and the other put 5%.
A size rule protects you from the outcome you did not expect. A position that is small enough cannot sink the portfolio even if the thesis is wrong, which is what makes it possible to hold through volatility.
There is no legal limit for individuals, but there are reference points. US mutual funds that call themselves diversified must meet the 75-5-10 test in the Investment Company Act of 1940, section 5(b)(1).
| Rule | What it says | Note |
|---|---|---|
| 75-5-10 (diversified funds) | At least 75% of assets in cash, government securities, other funds, or positions of no more than 5% of assets and 10% of the issuer’s voting shares | The other 25% has no per-issuer cap |
| Fixed cap (illustrative) | No single stock above a set share, for example 5% or 10% | Simple; ignores how volatile each stock is |
| Volatility sizing | Weight ∝ 1 ÷ volatility | Equalizes stand-alone risk |
| Crash budget | Weight ≤ accepted loss ÷ worst past drop | Ties size to a loss you can live with |
| Kelly fraction | Bet size from edge and odds | Assumes you know your edge; most use a fraction of it |
The Kelly criterion entry explains why full Kelly sizing is too aggressive when the edge is estimated. The concentration side is covered in single-stock concentration limits.
Give each holding the same amount of stand-alone risk. Pick a risk allowance per position, then divide it by the holding’s volatility. With an illustrative allowance of 3 points of volatility per position, using 5-year monthly volatility from Exent’s data as of Sep 30, 2026:
| Holding | 5-year volatility | Weight for 3 points of risk | Inverse-volatility weight (five stocks) |
|---|---|---|---|
| NVDA (Nvidia) | 50.1% | 6.0% | 10.2% |
| XOM (Exxon Mobil) | 27.0% | 11.1% | 18.9% |
| AAPL (Apple) | 24.7% | 12.1% | 20.7% |
| JPM (JPMorgan Chase) | 24.1% | 12.4% | 21.2% |
| JNJ (Johnson & Johnson) | 17.6% | 17.0% | 29.0% |
The last column scales the same idea so five stocks fill 100%. Volatility sizing ignores how holdings move together; two calm stocks in the same industry still form one cluster. The weighting rules are compared in how much of each stock to own.
Decide the most you would accept losing from one holding in a crash, as a share of the whole portfolio. Then divide by the worst drop that holding has had. With an illustrative 5% budget and peak-to-trough drops from the same data:
| Holding | Drop in 2022 bear market | Max weight (5% budget, 2022) | Worst drop in five crises | Max weight (worst crisis) |
|---|---|---|---|---|
| NVDA | 61.8% | 8.1% | 85.1% (2008) | 5.9% |
| JPM | 39.5% | 12.7% | 68.1% (2008) | 7.3% |
| AAPL | 28.5% | 17.5% | 60.9% (2008) | 8.2% |
| XOM | 20.5% | 24.4% | 47.9% (2020) | 10.4% |
| JNJ | 13.9% | 36.0% | 35.5% (2008) | 14.1% |
| GBTC (bitcoin fund) | 68.3% | 7.3% | 68.3% (2022) | 7.3% |
The crisis you choose changes the answer a lot. A stock that held up in 2022 can still have fallen hard in 2008. GBTC’s own 5-year maximum drawdown was 83.3%, which would cap it at 6.0%. This method works for single positions; a broad index fund is the core the budget protects, not a position to size this way.
Compare each holding’s share of past crash losses with its weight. One name for this is crash liability: each holding’s share of the portfolio’s loss in its worst 5% of months. A holding whose share is far above its weight is oversized for the risk it brings.
A simpler version replays a named crisis. In a sample portfolio of 40% SPY, 25% QQQ, 20% Apple and 15% Nvidia, replaying each holding’s 2022 drop, Nvidia caused 27% of the loss: about 1.8 times its weight. Try it with the crash liability tool.
More than most rules of thumb allow. In an anonymized sample of 144 portfolios analyzed in Exent (internal accounts excluded), the median portfolio had 15 holdings, the median largest holding was 26.1% of the money, and the median top five were 77.2%. In 77.1% of portfolios the top five were more than half the money. Concentration is not wrong in itself; it is a bet that should be sized on purpose. See concentration risk.
More methods are under portfolio optimization.
There is no universal number. Diversified US funds keep at least 75% of assets in cash, government securities, other funds or positions of 5% or less, and many individual investors use a fixed cap, but the size that fits depends on the stock’s volatility and the loss you can accept.
It usually refers to keeping any single stock at or below 5% of a portfolio. It mirrors the 75-5-10 test that US diversified funds must meet under the Investment Company Act of 1940.
Divide a risk allowance by the holding’s volatility, or divide the loss you accept from that holding by its worst past drawdown. Use the stricter result.
It depends on the stock. A 20% position in a stock that fell 60% in a crash would cost 12% of the whole portfolio, so compare that loss with what you could hold through.
Broad index funds are already diversified across hundreds of companies, so they are usually treated as the core rather than sized like single stocks. Watch overlap with your single stocks, which are often inside the fund.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.