How Much Should I Put in One Stock? Position Sizing Explained
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.
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.
What is your own portfolio exposed to? Check it with Exent →
They pick a weighting rule, then adjust for conviction and limits. A rule matters because without one, weights are set by the order you bought things and by which ones went up. Four rules cover most of practice:
| Method | Rule | Needs | Main weakness |
|---|---|---|---|
| Equal weight | Same dollars in each holding | Nothing | Volatile holdings dominate the risk |
| Inverse volatility | Weight ∝ 1 ÷ volatility | Each holding’s volatility | Ignores correlations |
| Mean-variance | Maximize return per unit of risk | Expected returns, volatilities, correlations | Very sensitive to return guesses |
| Risk budgeting | Each holding gets a set share of risk | Volatilities and correlations | Favors low-volatility assets; can need leverage to hit a return target |
Sample portfolio: Nvidia, Apple, JPMorgan Chase, Johnson & Johnson and Exxon Mobil. Volatility is each stock’s 5-year monthly volatility in Exent’s data as of Sep 30, 2026. Portfolio volatility and risk shares are estimates from a one-factor model (each stock’s link to the S&P 500 plus its own risk).
| Stock | 5-year volatility | Equal weight | Inverse volatility | Equal risk |
|---|---|---|---|---|
| NVDA | 50.1% | 20% | 10.2% | 8.8% |
| AAPL | 24.7% | 20% | 20.7% | 18.0% |
| JPM | 24.1% | 20% | 21.2% | 18.5% |
| JNJ | 17.6% | 20% | 29.0% | 32.6% |
| XOM | 27.0% | 20% | 18.9% | 22.1% |
| Estimated portfolio volatility | 18.4% | 15.4% | 14.7% | |
| Portfolio beta to S&P 500 | 0.95 | 0.77 | 0.69 |
The same five names give three quite different portfolios. Equal risk puts a third of the money in the calmest stock and less than a tenth in the most volatile one.
It is simple and avoids over-trusting forecasts, but equal dollars are not equal risk. In the sample, Nvidia at 20% of the money carried an estimated 46% of the volatility, while Johnson & Johnson at 20% carried about 6%. See equal-weighted.
At index level, equal weighting has had mixed results. In a backtest from July 2016 to September 2026, the equal-weight S&P 500 (RSP) returned 11.5% a year with a Sharpe of 0.62, against 15.2% and 0.88 for the cap-weighted index, a decade in which the largest companies led. The comparison is on the equal-weight S&P 500 page.
Divide 1 by each holding’s volatility, then scale the results to add up to 100%. A stock half as volatile gets twice the weight. It needs no return forecast, which is its main appeal. It ignores correlations, so two calm stocks that always move together are treated as two separate bets. More in inverse-volatility weighting.
A risk parity (inverse volatility) backtest weights SPY, TLT and GLD by their 12-month volatility. From July 2017 to September 2026 it returned 8.7% a year with 10.4% volatility and a Sharpe of 0.65, against 14.9%, 15.9% and 0.82 for the S&P 500. It still fell 18.2% in 2022, because its bond sleeve fell with stocks.
It is the textbook answer and the hardest to use well. Mean-variance optimization finds the weights on the efficient frontier, but it treats your return estimates as facts. Small changes in those estimates swing the weights to extremes. The backtests below re-estimated the inputs every month from the trailing 36 months, across SPY, QQQ, IWM, TLT, GLD and AGG:
| Backtest, Jul 2019–Sep 2026 | Annual return | Volatility | Sharpe | Max drawdown |
|---|---|---|---|---|
| Max-Sharpe optimizer | 13.5% | 13.6% | 0.83 | 30.5% |
| Minimum-variance optimizer | 0.7% | 5.9% | −0.27 | 17.1% |
| S&P 500 (SPY) | 15.7% | 16.3% | 0.83 | 23.9% |
| Backtest, Jul 2019–Sep 2026 | Max-Sharpe optimizer | Minimum-variance optimizer | S&P 500 (SPY) |
|---|---|---|---|
| Annual return | 13.5% | 0.7% | 15.7% |
| Volatility | 13.6% | 5.9% | 16.3% |
| Sharpe | 0.83 | −0.27 | 0.83 |
| Max drawdown | 30.5% | 17.1% | 23.9% |
The max-Sharpe version ended with the same Sharpe as the index and a deeper drawdown, losing 27.8% in 2022. Pitfalls are covered in optimization pitfalls.
You decide each holding’s share of total risk first, then solve for the weights that deliver it. The equal-risk version gives every holding the same share; in the sample, that is 20% of the volatility each. A risk budget can also be unequal, for example a larger share for a core index fund. It uses risk contribution, which accounts for correlation, so it fixes the main blind spot of inverse volatility.
The cost is a tilt toward calm assets. In the sample, Nvidia had returned 61.7% a year over five years and Johnson & Johnson 13.4%, so the weights that cut risk most would have cut return too. Risk budgeting controls risk; it does not know which stock will lead.
Size limits are covered in position sizing. The full topic sits under portfolio optimization.
Equal weights are simple but not equal risk. In a sample five-stock portfolio, Nvidia at 20% of the money carried an estimated 46% of the volatility.
Inverse volatility weights each holding by 1 divided by its volatility and ignores correlations. Full risk parity solves for weights that give each holding the same share of total risk, correlations included.
It is sensitive to return estimates, which are noisy. In Exent’s backtest from mid-2019, a max-Sharpe optimizer matched the S&P 500’s Sharpe ratio and had a deeper maximum drawdown.
Divide each holding’s market value by the portfolio’s total value, including cash. Then compare those weights with each holding’s share of risk, which is often very different.
No. The examples show how weighting rules work on a sample portfolio with past data. The right weights depend on your goals, horizon and situation.
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.