Position Sizing: Optimal Investment Allocation

Position sizing answers a simple but critical question: how much of your portfolio should you allocate to each investment? Getting this right often matters more than finding the perfect stock.

What you'll learn

The core idea of position sizing and why it drives long-term outcomes
Fixed fractional, fixed risk per trade, and volatility targeting methods
Kelly criterion and fractional Kelly for growth-efficient sizing
Risk parity and covariance-aware sizing across multiple assets
How drawdowns, risk of ruin, and constraints shape real-world sizing
Step-by-step calculations with concrete examples
Practical workflows and checks professionals use before allocating

Concept explanation

Position sizing is the process of deciding how many dollars or shares to allocate to a position given your goals and constraints. Instead of asking only is this a good investment, you also ask how big should it be. Two investors can buy the same asset and end up with very different results because they sized their positions differently.

At its heart, sizing is a trade-off between return and risk. Bigger positions amplify both gains and losses. Smaller positions lower risk but can make good ideas irrelevant in the portfolio. The goal is to find a size that aligns with your risk tolerance, drawdown limits, and the characteristics of the asset, such as volatility and correlation with your other holdings.

Professionals combine simple rules of thumb with statistical models. They might start with a risk budget per position, scale by volatility so each idea contributes similar risk, and then adjust for correlation so the portfolio does not unintentionally concentrate on one risk factor. They also account for uncertainty in estimates by using conservative or fractional versions of aggressive formulas like Kelly.

Why it matters

Position sizing is one of the largest drivers of long-term compounding. Even with a solid win rate, oversizing can lead to deep drawdowns that are hard to recover from mathematically. A 50 percent loss requires a 100 percent gain to break even. Sizing helps cap losses to preserve the capital you need for future opportunities.

Sizing is also how you express conviction and risk budgets. If two ideas have similar expected returns but one is twice as volatile, equal dollar allocations will not deliver equal risk. Without sizing, your portfolio can become dominated by the most volatile assets. Proper position sizing helps normalize risk outcomes and makes performance more consistent.

Calculation method

Below are common methods used in practice, from simple to advanced. You can combine them.

Fixed fractional or percent-of-equity

Idea: Risk a constant fraction of your portfolio on each position or trade.
Setup: Decide a risk budget per position, for example 1 percent of account value.
If using a stop-loss, shares equal risk budget divided by per-share risk.

Example

Account 25,000, risk budget 1 percent equals 250.
Stock price 50, stop at 45, per-share risk 5.
Size equals 250 divided by 5 equals 50 shares, or 2,500 dollars notional.

Volatility targeting

Idea: Size positions so each contributes a similar volatility to the portfolio.
Rule of thumb: Weight is proportional to target volatility divided by asset volatility.

w_i = \begin{cases} \text{Dollar weight} = \text{TargetVol} / \sigma_i, \\ \text{Position size in shares} = (\text{Dollar weight}) / \text{Price} \end{cases}

Example

Target annualized vol 10 percent.
Asset estimated vol 20 percent. Weight equals 10 divided by 20 equals 0.5 of capital.
On a 100,000 account, allocate 50,000.

Tip: Many traders use ATR Average True Range as a volatility proxy on daily data. For a futures or FX contract, you can convert ATR to dollar risk per contract and back out the number of contracts.

Fixed risk per trade using ATR

Decide per-position risk budget, for example 0.5 percent to 2 percent.
Compute per-share risk as k times ATR, where k might be 2 times.
Shares equal risk budget divided by per-share risk.

\text{Shares} = \frac{\text{RiskBudget}}{k \times ATR \times \text{DollarVolPerPoint}}

Kelly criterion for growth-efficient sizing

Single bet binary payoff version:

f^* = \frac{bp - q}{b}

Where b is the net odds per unit staked, p is probability of win, q is 1 minus p.

For continuous returns with mean excess return mu and variance sigma squared, a common approximation for the optimal fraction is:

f^* \approx \frac{\mu}{\sigma^2}

For multiple assets, the Kelly weights are proportional to the inverse covariance times expected excess returns:

\mathbf{w}_{Kelly} = \Sigma^{-1} \mu

Caution: Full Kelly is very aggressive and sensitive to estimation error. Many practitioners use fractional Kelly, for example 0.25 to 0.5 times the Kelly result.

Risk parity across multiple assets

Idea: Choose weights so each asset contributes equally to total portfolio volatility.
Given covariance matrix Sigma and weight vector w, total portfolio volatility is sqrt of w transpose Sigma w. The risk contribution of asset i is weight i times the i-th element of Sigma times w, divided by total volatility.

RC_i = \frac{w_i (\Sigma w)_i}{\sqrt{w^\top \Sigma w}}

Risk parity seeks RC i equal to total risk divided by number of assets, often solved numerically with constraints such as no shorting and weight bounds.

Drawdown and risk-of-ruin constraints

You can back into a risk budget using the number of trades N, win rate p, payoff ratio R average win divided by average loss, and maximum acceptable drawdown D.
Approximate maximum drawdown scales with risk per trade and losing streak length. The probability of a losing streak of length L in N trials with win rate p is roughly one minus the cumulative probability of all runs shorter than L. Practitioners often simulate or use conservative tables to choose risk per trade so that likely drawdowns stay under D.

Constraint-aware sizing

Incorporate limits: maximum weight per position, sector caps, liquidity minimums, turnover, taxes, and borrowing or margin limits.
Solve as an optimization problem with objective such as risk parity or mean-variance utility and constraints such as sum of weights equals 1, weights greater than or equal to 0, and individual caps.

Case study

Scenario

Portfolio size: 100,000.
Universe: three ETFs. Equity E with annualized vol 18 percent, Bond B with vol 6 percent, Commodity C with vol 15 percent.
Estimated correlations: E with B equals 0.1, E with C equals 0.3, B with C equals 0.0.
Target portfolio volatility: 8 percent.

Step 1: Volatility targeting baseline

Uncorrelated approximation weights proportional to 1 divided by vol.
Raw weights proportional to 1 divided by 18 percent, 1 divided by 6 percent, 1 divided by 15 percent equals 0.0556, 0.1667, 0.0667.
Normalize so they sum to 1: weights equal 0.19 E, 0.57 B, 0.23 C.

Step 2: Adjust for correlation and scale to target vol

Compute portfolio vol with these weights using Sigma from inputs. Using the given correlations and volatilities, the resulting vol is approximately 7.9 percent, close to the 8 percent target.
Dollar allocation on 100,000 equals 19,000 E, 57,000 B, 23,000 C.

Step 3: Add maximum position cap and liquidity check

Suppose policy says max 50 percent per asset and minimum average daily dollar volume threshold is met for all three ETFs. B at 57 percent breaches the cap.
Reallocate the 7 percent excess from B proportionally to E and C by their risk weights. The new weights approximate 0.23 E, 0.50 B, 0.27 C. Recompute portfolio vol, now near 8.1 percent. Acceptable.

Step 4: Incorporate expected returns via fractional Kelly tilt

Assume expected annual excess returns: E equals 5 percent, B equals 1 percent, C equals 3 percent.
Compute Kelly vector Sigma inverse times mu and then take 0.3 times to apply 30 percent Kelly. This produces a modest tilt toward E and C.
Blend with risk parity base using 70 percent weight on risk parity and 30 percent on fractional Kelly to form final weights.

Result

A balanced, volatility-targeted portfolio respecting caps, with a modest return tilt and target 8 percent vol.

Practical applications

Equity selection with stops: Use fixed risk per trade tied to ATR. If ATR is 2 dollars, stop at 2 times ATR equals 4 dollars below entry, and you risk 200 on a 25,000 account 0.8 percent, then shares equal 200 divided by 4 equals 50.
ETF rotation: Use volatility targeting to equalize risk contribution across chosen ETFs, then cap concentrations and adjust for correlations monthly.
Options: Convert option delta and underlying ATR into dollar risk. If delta is 0.5 and underlying ATR is 3 dollars, per-contract exposure is roughly 0.5 times 3 times 100 equals 150 dollars. Size so risk per position stays under your budget.
Futures: Use contract specifications to convert ATR points to dollars. Contracts often have high notional exposure, so volatility targeting and risk per trade both help avoid oversizing.
Income portfolios: Use drawdown limits to back into per-position risk. If your maximum drawdown tolerance is 10 percent, you can set per-position risk so that even a cluster of losses remains within that limit.
Multi-asset strategies: Use risk parity or minimum variance to balance risks. For a return tilt, blend in fractional Kelly or mean-variance optimization with conservative expected returns.

Common misconceptions

よくある誤解

- Position sizing is only for traders with stop-losses. In reality, long-term investors also size positions to control drawdowns and diversify risk. - Equal dollar weights mean equal risk. More volatile assets dominate risk unless you adjust sizes. - Kelly is a free lunch for maximum growth. Full Kelly is highly sensitive to errors and can cause severe drawdowns. Fractional Kelly is more practical. - Volatility targeting ignores return. You can blend risk-based sizing with expected return tilts while respecting risk budgets. - Correlations are stable. Correlations change in stress periods and often rise toward one. Re-estimate and stress test regularly.

Advanced considerations

Estimation error: Expected returns and covariance estimates are noisy. Use robust methods such as shrinkage covariance, Bayesian or Black-Litterman adjustments, and regularization or weight bounds.
Regime shifts: Volatility and correlation jump in crises. Use regime-switching vol estimates, long and short lookbacks, and stress scenarios to avoid oversizing.
Transaction costs and slippage: Include costs in your optimization by penalizing turnover or using a costs-aware objective. Consider minimum trade sizes to avoid excessive churn.
Liquidity and capacity: Cap positions by a percent of average daily volume and by expected days to exit. Sizing should drop as liquidity falls.
Leverage and margin: Vol targeting can imply leverage for low-vol assets. Check margin requirements, financing costs, and potential for volatility spikes.
Taxes: Rebalancing frequency affects taxable gains. Consider wider bands or thresholds to reduce turnover.
Discrete shares and minimum trade sizes: Round sizes while staying under risk budgets. If rounding pushes you over a risk cap, reduce, do not increase, the position.

Summary

まとめ

- Position sizing controls risk and drives long-term compounding more than security selection alone. - Simple methods include fixed fractional risk and ATR-based sizing for clear stop levels. - Volatility targeting equalizes risk contribution and avoids concentration in volatile assets. - Kelly and fractional Kelly link size to expected edge but require conservative use. - Risk parity and covariance-aware methods balance risks across multiple assets. - Constraints such as caps, liquidity, and taxes must shape real-world sizes. - Re-estimate vol and correlation, stress test, and use fractional or blended approaches to handle uncertainty.

Glossary

ATR: Average True Range, a measure of recent price volatility used to set stop distances and risk.

Volatility targeting: Sizing positions so each contributes a similar amount of volatility to the portfolio.

Kelly criterion: A formula that gives the theoretically optimal fraction of capital to wager for maximum growth.

Risk parity: An allocation where each asset contributes equally to total portfolio risk.

Covariance matrix: A matrix capturing variances and pairwise covariances of asset returns, used to model portfolio risk.

Fractional Kelly: Using a fraction of the Kelly size to reduce drawdowns and estimation risk.

Risk of ruin: The probability that losses reduce capital below an unacceptable threshold.