Position sizing is the only edge you fully control

Most traders spend their attention on entries, some on exits, and almost none on size. The ordering is exactly backwards relative to how much of your outcome each decision explains. Entry timing is largely a bet on a distribution you do not control. Size is a decision you make unilaterally, in advance, with complete information about your own account. It is the one input the market cannot argue with.

Start with a simple, unglamorous rule: fixed fractional risk. Decide what percentage of equity a single losing trade may cost — for most people somewhere between 0.25% and 1% — then let the distance to your invalidation level determine the number of shares or contracts. A wide stop gets a small position, a tight stop gets a larger one, and every loss costs approximately the same. That last property is what keeps a bad streak survivable rather than terminal.

The arithmetic of drawdown is why this matters more than any signal. A 20% loss requires a 25% gain to recover. A 50% loss requires 100%. A 70% loss requires 233%, which in practice means the account is over. Sizing rules are not conservatism for its own sake; they are what keeps you inside the region where recovery is arithmetically plausible.

The refinement worth making next is volatility scaling. Risking 0.5% of equity on an instrument whose daily range has tripled is not the same bet it was last month, even though the percentage is unchanged. Dividing your intended risk by a recent measure of the instrument’s volatility — average true range works fine — keeps the economic size of the bet roughly constant as conditions change. It also automatically shrinks you in exactly the environments that punish size.

Then there is correlation, which is where most retail portfolios quietly break. Six positions at 0.5% risk each look like 3% of exposure. If all six are semiconductor names in the same week, it is one position of roughly 3% wearing six hats. Cap correlated exposure explicitly: group by sector, factor or realised correlation, and treat the group as the unit that has a limit.

Kelly sizing comes up whenever this topic does, and it deserves one paragraph of caution. Full Kelly maximises long-run growth only if you know your edge exactly, which you never do. Estimation error in the edge translates into catastrophic overbetting. Practitioners who use it at all use a quarter or a half, and they treat the result as a ceiling rather than a target.

Write your rules down before the session, in numbers rather than adjectives, and make them mechanical enough that a spreadsheet could apply them. The purpose is not to be strict for its own sake. It is that the version of you deciding size at 09:45 on the third losing day in a row is not the version who should be making that call.