Every trading question eventually reduces to this one, because the answer decides whether you are still around for the others: when you put on a position, what fraction of your account is on the line if it goes straight to your stop?
The standard: fixed-fractional risk
The professional convention is fixed-fractional: risk the same percentage of current equity on every trade — commonly 0.5% to 2% for discretionary traders, with 1% the default worth earning your way off. "Risk" means the actual loss at your stop: a $10,000 account risking 1% loses $100 if the stop is hit, whatever the position's face size. Position size is then derived, never chosen: size = (account × risk%) ÷ stop distance. A 50-pip stop on that account means $2 per pip; a 20-pip stop means $5 per pip. The stop's location comes from the chart; the dollars come from the formula; the size falls out at the end. Traders who pick size first and then look for somewhere to put the stop have the pipeline backwards.
Why so small? The survival math
Losing streaks are not a possibility — they are an arithmetic certainty. A system that wins 50% of the time will produce a streak of seven straight losses roughly once every 128 sequences, which for an active trader means several times a year. Now compare survival across risk levels: at 1% per trade, ten straight losses draw you down about 9.6%; at 5%, the same streak costs ~40%; at 10%, ~65%. And recovery is asymmetric — a 10% drawdown needs 11% to get back, but 50% needs 100%. Small per-trade risk is not timidity; it is what makes an ordinary bad month mathematically boring instead of terminal.
The same math exposes the beginner's actual failure mode: it is rarely one bad trade. It is 5% risk, three correlated positions, and a losing week — a perfectly normal cluster of outcomes that fixed-fractional sizing at 1% would have shrugged off.
Adjustments worth making — and not
Scale with equity, automatically. Fixed-fractional does this by construction: risk shrinks in drawdowns (defense exactly when your edge is in doubt) and grows with profits. Recompute from current equity, not starting equity.
Count themes, not tickets. Three 1% trades that are all effectively short the dollar are one 3% trade. Cap total risk-at-once (many traders use 3–5% across all open positions) and treat correlated positions as one line item.
Grade your setups, honestly. Risking more on "A+" setups is defensible only after your journal proves your A+ grades actually win more — most beginners' confidence is uncorrelated with their outcomes. Until the data exists, flat 1% removes the self-deception channel entirely.
Skip martingale, permanently. Doubling risk after losses to "win it back" converts a losing streak into account death on a schedule. Any sizing rule that grows risk as equity falls has the survival math exactly inverted.
Setting your own number
Start from the drawdown you can genuinely tolerate — not financially, psychologically: the level at which you know you would start revenge-trading or freeze. Suppose that is 15%. At 1% risk, reaching it takes roughly fifteen net losses — a bad but survivable stretch; at 3%, five. Pick the per-trade fraction whose plausible losing streak stays inside your real tolerance, then subtract a little, because everyone overestimates their tolerance before the streak actually happens. New traders and new systems belong at 0.25–0.5% — the tuition period should be cheap. The percentage sounds like the least interesting number in trading; it is the one doing the most work.