Position Sizing

Starter — for Never bought a share

Position sizing is deciding how much money to put into one trade, before you know how it turns out. It is the main thing that separates a survivable account from a fragile one.

Why size matters as much as the idea

Two traders can both be right 50% of the time and one of them can go broke. The difference is how much they risked on each attempt. If losing a trade costs 1% of the account, ten losses in a row still leaves you with most of your money. If each loss costs 20%, five bad trades in a row ends the account.

The size follows the stop

Choose where you will admit you were wrong — the price at which the reason you bought no longer holds. The distance from your entry to that price is your risk per share. Divide the money you are willing to lose by that distance and you have the number of shares. A wider stop means a smaller position, not a bigger loss.

A common starting rule

Many traders start with 1% of the account at risk per trade. On a $50,000 account that is $500 of potential loss, whatever the position turns out to be worth. The rule is not magic — it is a discipline that keeps a normal losing streak from becoming a crisis while you learn.

What to take away

• Decide how much you can lose before you decide how much you can make.
• Size = risk budget ÷ distance to the stop.
• A 1% risk-per-trade rule keeps a losing streak survivable.

Medium — for Invested before, reads the news

Risk-based sizing converts a dollar risk budget and a stop distance into a share count, so every position risks the same amount regardless of volatility or conviction.

The arithmetic and why it is volatility-normalizing

Shares = (account × risk %) ÷ (entry − stop). A high-volatility name has a wider stop, so it gets fewer shares — and each position contributes roughly the same dollars of risk to the portfolio. This is the core reason the formula works: it makes the account's sensitivity to being wrong constant, even though the trades are not.

Streaks are the reason for the rule

At a 50% win rate, the chance of five losses in a row across 100 trades is high enough to be routine, and longer streaks are not rare. At 1% risk, a ten-loss streak costs under 10% of the account. At 5% risk, the same streak costs 40% — and needs a 67% gain to recover. The rule is a probability statement about how often normal luck looks like a disaster.

Portfolio-level limits

Per-trade risk is only the first constraint. Others: total open risk across all positions; a cap on any one position as a percentage of the account; a sector or factor cap, because correlated positions are one position wearing several names; and a drawdown circuit-breaker that reduces size after a defined loss run. These are the limits that stop a bad week from becoming a bad year.

Where sizing is chosen instead

For buy-and-hold portfolios, size is allocation — the fraction of the portfolio in each asset — and the volatility-normalizing equivalent is a risk-parity or minimum-variance weighting rather than equal dollars. For options with defined risk, the maximum loss is known in advance, so the sizing arithmetic simplifies to how many contracts at that known loss. The principle is unchanged: decide the loss first, then the size.

What to take away

• Size = risk budget ÷ stop distance normalizes risk across trades of different volatility.
• At 1% risk, long losing streaks are survivable; at 5%, they are the account.
• Add portfolio caps — total open risk, single-position cap, sector cap, drawdown circuit-breaker.
• For allocation portfolios the same logic appears as weighting; for defined-risk options the max loss is known up front.

High — for Works with this number already

Position sizing is the mapping from an uncertain edge to a survival probability; the professional version derives the risk per trade from the strategy's own loss distribution rather than from a fixed percentage.

From the loss distribution, not a convention

The fixed 1% rule is a reasonable default, not a derived answer. The derived version starts from the strategy's statistics: win rate, average win/average loss, the dispersion of outcomes, and the observed worst run. Given those, you can choose the risk fraction that keeps the probability of a ruinous drawdown below a threshold you name — usually expressed as a risk of losing a set fraction of capital within a horizon. Kelly-style sizing appears here as an upper bound, not a target: full Kelly is maximally aggressive for a known edge, and edges are estimated with error, so fractional sizing is the norm.

Volatility scaling and correlation

When trades are held simultaneously, the relevant quantity is portfolio risk, not per-trade risk. Two useful adjustments: scale each position inversely to its own volatility (so the risk contributions are comparable), and apply a correlation haircut — a cluster of positions with high pairwise correlation should be sized as one exposure. Equity factor exposure does this implicitly, which is why a book of five "different" momentum names can behave like a single leveraged bet on the factor in a deleveraging.

Stops, gaps and the stop that does not fill

A stop is an instruction, not a guarantee. Overnight gaps, halted names and illiquid strikes can fill far below the stop, so realized loss per trade exceeds planned loss per trade with some frequency. Robust sizing budgets for that: use a conservative stop distance, assume slippage, and either reduce size for names with gap history (earnings, biotech, low liquidity) or express the position in a defined-risk structure — a long option, a spread, or a short call against stock — where the worst case is contractual.

Sizing the whole book against the objective

The account-level question is the one that matters: what gross and net exposure, what cash buffer, what drawdown tolerance? For a person, the binding constraint is behavioural — size to the level at which you would still follow the plan after three losses, because an optimally-sized position you abandon is worse than a slightly small one you keep. For a fund, the constraint is contractual, and the risk budget is allocated top-down across strategies with a correlation-aware framework (risk parity, equal risk contribution, or a formal risk-factor budget).

The failure modes that actually destroy accounts

Averaging down a losing position without re-deriving size, so the original risk budget is quietly doubled or tripled. Letting a winner's position grow past the cap and then treating it as a core holding. Increasing size after a losing streak to "make it back". Sizing on margin so a normal gap forces liquidation at the worst price. Every one of these is a sizing decision taken after the outcome was partly known — which is exactly what the discipline exists to prevent.

What to take away

• Derive risk per trade from the strategy's loss distribution and a stated maximum-drawdown tolerance; 1% is a default, not a law.
• Treat highly correlated positions as one exposure and scale sizes inversely to volatility.
• Budget for stops that do not fill — gaps, halts and thin liquidity — or use defined-risk structures.
• Size to what you will still hold after three losses; the binding constraint is behavioural, and it is knowable in advance.