This article explains sizing arithmetic with worked examples. The numbers are illustrations of a method, not recommendations of any risk level, position, or strategy. Nothing here is investment advice. Decisions are yours.
Position sizing answers the only question a trader fully controls: how much do I lose if I am wrong? In a market where ordinary daily noise runs 4-8% and wicks run double that, the equity-market habit of sizing by conviction, or by round percentages of the account, imports a fatal assumption. Here is the arithmetic that replaces it.
Risk-first sizing, the method
The professional convention inverts the amateur question. Amateurs ask "how much should I buy"; the method asks "how much am I willing to lose on this idea," fixes that number first, and derives the position from it. The formula is one line: position size equals risk budget divided by stop distance. Define the risk budget as a fraction of the account, the classic range being fractions of one percent to low single digits per idea; define the stop distance as the percentage from entry to invalidation, per the geometry work in the stop-placement piece; divide. Everything else in sizing is commentary on those two inputs.
The worked example, crypto edition
Take a $20,000 account risking 1% per idea, $200, on an alt whose 14-day ATR runs 8%. A stop tight enough for an equity, say 5%, sits inside this asset's daily noise and will be collected by an ordinary wick; suppose the honest invalidation, derived from the asset's actual excursion behavior, sits 16% below entry. Position size: $200 divided by 0.16, equals $1,250, about 6% of the account. Now the instructive contrast: the same $200 risk on a quiet major with a 6% stop sizes to $3,333, nearly triple the position. Same account, same risk, radically different exposures, and that is the method working: VOLATILITY SHRINKS SIZE. The amateur pattern runs exactly backwards, oversized positions in the wildest assets because the upside excites, with stops tightened to pretend the risk away, which converts one honest 16% invalidation into four consecutive 5% wick collections. Same losses, no information, and the idea was never actually tested.
Portfolio heat, and the correlation trap
Per-idea risk compounds into portfolio heat: five open positions at 1% each is 5% of the account at risk if everything invalidates. Equity intuition treats that as conservative because equity positions fail semi-independently. Crypto's one-factor structure voids the assumption: when BTC turns hard, alt correlations converge toward one and ALL stops trigger together, per the stress behavior documented in the sectors piece, so portfolio heat should be read as a SINGLE bet's risk, not a diversified sum. Our founding experiment paid for this lesson in public: four concurrent 20% positions, modest by equity standards, behaved as one leveraged BTC position and drew down 74%, per the transplant experiment. The practical correction is a total-heat ceiling set as if correlations were one, because in the moments that decide survival, they are.
The survival arithmetic underneath
Why the fractions stay small: losing streaks are not tail events for any real strategy, they are scheduled. A system winning 40% of the time, respectable for trend approaches treated in the breakout base rates, will produce eight-loss streaks routinely across a few hundred trades. At 1% risk per idea, eight straight losses cost roughly 8% of the account, an annoyance. At 5%, the streak costs a third of the account, entering the recovery curve's punitive region, covered in the drawdown history. At 10% with leverage, the streak is terminal, and the market schedules the streak regardless of anyone's conviction. Sizing is the instrument that converts an expectancy edge into survival long enough to collect it; without it, the same edge is just a countdown with better marketing.
The habits that make it real
Four, each mechanical. Fix the risk fraction in writing before the season heats up, because sizing discipline is exactly what euphoria dissolves, per the bull-market-genius piece. Derive stops from the asset's measured behavior, never from the size you wish you could hold; if the honest stop makes the position embarrassingly small, that is the method protecting you from the asset, not from the opportunity. Cap total heat as one correlated bet. And size DOWN in deteriorating regimes rather than doubling to recover, because the get-back-to-even reflex is the documented mechanism by which recoverable drawdowns become terminal ones. None of this predicts anything, which is the point: sizing is the part of trading that works identically whether any given idea does. Decisions are yours.