Position Sizing
Position sizing is the decision of how many units of something to buy or sell in a single trade, based on how much money you're willing to risk rather than how much you simply feel like buying. Say a trader has a $10,000 account and wants to risk 1% ($100) on a gold trade. Gold futures move in dollar amounts per contract, so if a stop-loss 10 points away would cost $100 per contract at that size, the trader can only take one contract - not three, even if they're confident, because three contracts would put $300 at risk on a single trade.
Getting this step right is what separates a bad prediction from a career-ending one. A trader who sizes positions correctly can be wrong on a trade and lose a small, predictable amount; a trader who sizes by gut feeling can be right most of the time and still blow up the account on the one trade sized too large when it goes wrong. The calculation - account size, percent risked, distance to stop-loss - takes under a minute and should happen before every trade, not as an afterthought.
The mechanics shift by market but the logic doesn't. In forex, position size is usually expressed in lots (a standard lot is 100,000 units of currency); in stocks, it's a share count; in crypto, it can be a fraction of one coin, since exchanges let traders buy 0.01 of a Bitcoin rather than requiring a whole unit. Whatever the unit, the question being answered is always the same: how much of this can I hold without one bad trade costing more than I decided in advance to accept?
The fixed-fractional method described above, risking a set percentage of the account per trade, is the right place to start and has one structural weakness: it holds the money at risk constant while the probability of hitting the stop varies enormously between instruments. A 30-pip stop is generous on a quiet currency pair and will be taken out by an ordinary hour on a volatile one. A trader using the same dollar risk on both is not running equal risk; they are running a system that loses far more often on the second instrument for reasons that have nothing to do with whether the idea was good.
The usual fix is to set the stop distance from the instrument's own volatility rather than from a round number, most commonly as a multiple of average true range, and then size the position so that distance costs the intended amount. Two trades placed this way carry equal money at risk and roughly equal odds of being stopped by noise, which is what equal risk should mean. The more aggressive alternative, sizing by the Kelly criterion to maximise long-run growth, is mathematically optimal only if the win rate and payoff ratio are known exactly, and they never are; estimating either slightly too high produces a size that maximises the chance of ruin instead. Practitioners who use it almost universally trade a half or a quarter of the recommended fraction, which tells you what they actually think of the estimates. This entry is paired with the one on position size, which covers the units the answer gets expressed in.