The stop loss is the only number in a trade you fully control. The entry depends on your fill, the target depends on the market's generosity, but the stop is yours alone, and it only does its job if it exists before the entry does. A stop chosen after you are in the trade is not a risk decision. It is a negotiation with a losing position, and the position usually wins.
The stop defines the trade
A stop is two things at once: your maximum loss and your statement of invalidation, the price at which the trade idea is provably wrong. Everything else derives from it. On a $10,000 account risking 1 percent, the budget is $100. A 20-pip stop on EURUSD at $10 per pip per standard lot allows 0.50 lots; a 40-pip stop allows 0.25. Same idea, same risk, half the size. The order of operations is fixed: stop first, size second, entry last. Traders who pick a size they like and then drag the stop until the numbers fit have run the process backward, and the account pays for it.
Structure-based stops
The default method: place the stop beyond the price that proves you wrong. For a pullback long in an uptrend, that is below the swing low that formed the pullback. Say EURUSD is trending up and you buy the retrace at 1.0850 with the swing low at 1.0832. The stop goes at 1.0826, six pips below the low to allow for spread and wick noise, for a 24-pip stop. If 1.0832 breaks decisively, the pullback thesis is dead, so getting stopped there is information, not bad luck. The same logic covers other setups: a range fade stops beyond the range extreme, a breakout trade stops on the far side of the base it broke from. The buffer deserves thought rather than habit: six pips suits EURUSD in London and is completely wrong for GBPJPY, where routine wicks run 15 pips, so scale it to the instrument's noise. And since short trades are stopped out at the ask, the spread effectively lives inside your buffer; widen it accordingly on anything with a fat spread.
ATR-based stops
Structure is not always clean, and momentum entries often trigger far from any obvious swing. The alternative is a volatility stop using Average True Range, the rolling average of recent bar ranges. If the 14-period ATR on the 15-minute EURUSD chart reads 12 pips during London, a stop of 1.5 to 2 times ATR, so 18 to 24 pips, sits outside routine noise by construction. On gold with a 15-minute ATR of $3.20 in the New York session, 1.5 times ATR gives a $4.80 stop; with $1 of movement worth $100 per lot, the $100 budget buys 0.20 lots. ATR stops adapt automatically: volatile days demand wider stops and therefore smaller size, which is exactly the adjustment most traders forget to make by hand.
Placement errors that feed the market
Three mistakes account for most stopped-then-reversed pain. First, parking stops at obvious round numbers or exactly at a swing point, where resting orders cluster; a buffer of a few pips or a fraction of ATR costs a little size and avoids the crowd. Second, stops tighter than the market's own noise: anything inside about one ATR of the entry is a donation, because normal fluctuation will find it regardless of direction. A tight stop only improves risk-reward if it still has a realistic chance of surviving. Third, the reversed process described earlier, where desired size chooses the stop. If an honest stop makes the position too small to bother with, the correct conclusion is that this trade is too expensive today, not that the stop should move.
One further habit separates tested traders from hopeful ones: the stop is written into the setup definition itself, so every trade of a given type uses the same placement logic. Standardized placement is what lets a journal answer real questions, such as whether 1.5 ATR outperforms 2 ATR on your breakout entries. At that point the stop is no longer a mood. It is a parameter with a track record.