A stop-loss is not an admission of weakness and not a formality the platform demands. It is the price at which your idea is objectively wrong, decided while you are calm, executed while you are not. Everything useful about risk management flows from taking that definition seriously.
Where stops belong
Behind invalidation, not behind comfort. If you bought a bounce off support, the trade is wrong when the support zone genuinely fails — so the stop goes beyond the zone, plus room for spread and noise. A stop placed at "the most I want to lose" instead of "where the idea fails" gets clipped by ordinary volatility and then watches the trade work without you. ATR helps here: if the market moves 80 pips on an average day, a 15-pip stop on a daily-chart idea is not a tight stop, it is a donation.
One unit to rule the ledger
Define R as the money risked on a trade — entry to stop, times pip value, times size. Then every outcome is a multiple: a win to a target twice the stop distance is +2R, a full loss is −1R, a breakeven scratch is 0R. The unit does two jobs. It makes trades of different sizes and pairs comparable, and it exposes the only statistic that matters: expectancy. A 40% win rate with average winners of +2R and losers of −1R earns +0.2R per trade — a losing hit-rate and a profitable system. Traders who never leave the percentage world keep chasing win rate; traders who think in R chase the ratio.
Managing the stop after entry
- Never widen a stop. Moving it away from price converts a defined loss into an open-ended one — the single most reliable account killer.
- Moving to breakeven early feels safe but strangles winners; many desks wait until price forms a new structural level to hide behind.
- Trail with structure (behind each new higher low) or with ATR — and journal every exit in R, so next month you can see which rule actually paid.
Your platform executes whatever rule you choose in two clicks. The rule itself has to exist before the trade does.