How to Size a Position

The decision that determines outcomes more than any view, made by almost everyone in the wrong order. Start from the loss you can accept and derive the size — never the reverse.

The inversion that fixes most of it

• Almost everyone decides how much to buy and then discovers what they can lose. The method is the other way round: decide the loss you can accept, then derive the size that produces it.
• Size is the only risk control that always works. Stops can gap, hedges can fail, correlations can converge — the number of units you own does none of those things.
• This page is arithmetic, not advice about how much anyone should risk. The percentages used are illustrative anchors, not recommendations.

1. Start from the loss, not the conviction

• Fix a share of capital you are prepared to lose on one position going wrong. Whatever number you choose, the discipline is that it is chosen before the position and does not move afterwards.
• Then define the exit — the level at which the thesis is wrong, in price terms. Not a round number, not a percentage picked for comfort: a level that means something about the position.
• The risk-based sizing calculator does exactly this arithmetic for any instrument.
• The failure mode: setting the exit close so the size can be large. That converts a sizing decision into a bet that ordinary noise will not touch your level — and ordinary noise usually does.

2. Size from volatility, not from price

• A €10,000 position in a 12%-volatility bond fund and a €10,000 position in a 60%-volatility cryptoasset are not the same position. They differ by a factor of five in the only dimension that matters.
• Volatility-scaled sizing equalises them: divide the target risk by each instrument's volatility. The result is that the "small" crypto position is genuinely small.
• Convert an annual volatility into the daily and horizon move it implies with the volatility converter. At 60% annualised, a 4% day is unremarkable — sizing that ignores this is sizing for a market that does not exist.
• Volatility changes. A position sized in a calm regime is oversized when volatility doubles, without you doing anything. This is what volatility targeting automates and what most people never revisit.

3. Leverage: size the exposure, not the deposit

• The deposit is not the position. A €500 margin deposit controlling €78,000 of index exposure is a €78,000 position, and every sizing calculation must use that number.
• Run any leveraged position through the margin-call simulator before opening it: the distance to the call is usually far shorter than intuition suggests.
• Leverage converts an ordinary mistake into a terminal one. The same wrong view at 1× is a bad year; at 20× it ends the position before you get the chance to be right — the pattern in LTCM and every case after it.
• Gaps defeat stops. Overnight and weekend moves skip the level entirely. Size assuming the stop may not work, because sometimes it will not.

4. Correlation: your positions are fewer than they look

• Five positions in the same sector are one position wearing five names. Total exposure to a risk factor is what matters, not the count of line items.
• The three-asset risk tool shows the gap that surprises people most: a 55% weight in the highest-correlation asset routinely carries 75–80% of portfolio risk.
• Stress the correlation assumption. Re-run any allocation with all pairwise correlations at 0.8. If the result is unacceptable, the portfolio depends on peacetime staying peaceful — see diversification.
• Crowding is a risk factor. A popular trade behaves differently when it reverses, because everyone exits at once — 2021 and Amaranth are the same lesson from opposite sides.

5. Sanity checks before committing

• The drawdown check. A 50% loss needs a 100% gain to recover. Run the drawdown calculator on the worst case and ask whether the recovery arithmetic is survivable.
• The Kelly check. Given an honest edge and odds, Kelly gives the growth-optimal stake. Almost nobody should trade full Kelly — the point of running it is that if your intended size exceeds it, the size is indefensible even on your own assumptions.
• The liquidity check. How many days of average volume is this position? If a full exit takes more than a day or two, your marks are optimistic and your stop is theoretical.
• The portfolio check. What does this position do to total risk, not to its own line? That is the stress test question.

The five common sizing errors

The checklist

• What loss am I accepting on this position, decided before I open it?
• Where is the exit, and is it a level that means something?
• What is the instrument's volatility, and have I sized to that rather than to price?
• What is the true exposure after leverage?
• What else in the portfolio moves with this, and what is the combined exposure?
• How long would a full exit take in a market that does not want to buy?