Stop placement, and the Nvidia gap that jumped a stop by $11.93

Nvidia closed Friday, Nov 6, at $186.64 and opened Monday at $173.07, about 7.3% lower, after its board said it would look into reports about its chief executive. Nothing traded in between during regular hours. For anyone holding a stop just under Friday's low, that weekend is the whole lesson: a stop decides when you get out, not the price you get.

What it is

A stop is the price at which a trade idea is proven wrong, written down before the trade starts. It turns a vague worry into a number, and it caps how long a losing idea can run. A stop order is the tool that enforces it: once the stock trades at or through the stop price, the order becomes a market order and sells at the next price available. That last detail is the one most people skip.

How it works

Placing a stop takes three checks. First, the chart: for a long, the natural spot is just under a level buyers defended, such as a recent low. Nvidia's Friday low was $185.22 at 12:50 PM ET, so a hypothetical stop at $185.00 sat just under it, $1.64 below the $186.64 close. Second, the noise: Nvidia's 14-day average true range through Friday was $8.02, about 4.3% of its price, which means a $1.64 stop sat inside about a fifth of one ordinary day. A stop that tight is betting the stock barely moves at all. Third, the order type: a stop order guarantees an exit once triggered but not the price; a stop-limit order, say a $185.00 stop with a $184.50 limit, guarantees the price but not the exit.

On this week's tape

Take the hypothetical long from Friday's $186.64 close with the $185.00 stop. Monday's first trade came at $173.07, so the stop triggered at the open and a market sale would have filled around there: $11.93 below the stop and $13.57 a share below entry, about 8 times the $1.64 the plan allowed. A wider stop two ATRs away, at $170.60, would not have escaped either, because Monday's low in the first 5-minute bar was $170.24. The stop-limit version shows the other trade-off. Monday's high was $179.92 at 10:30 AM ET and the stock closed at $178.67, so a $184.50 limit never filled; by Tuesday's low of $167.69 at 10:55 AM ET, that position was down about 10.2% from entry.

When it fails

Stops fail on gaps: earnings after the close, a weekend headline, a ruling or a data release before the open. Nvidia's weekend is the plain example, and both order types failed in their own way, one filling far below the stop and the other not filling at all. What limits the damage is size, not the stop: if the position was sized so that a gap several times the planned distance would still be survivable, the stop did its job of ending the trade even when it could not protect the price.

Common mistakes

The first mistake is treating the distance to the stop as the maximum loss, when it is only the expected loss in a market that trades continuously. The second is setting stops at round percentages, like 1% or 2%, without checking the stock's ATR; a stop well inside one day's normal range mostly measures noise. The third is moving a stop further away once price gets close, which turns a written plan back into a hope.

Try it on paper

Pick three stocks you follow and, for each, write down a recent swing low, the 14-day ATR and where a stop just under that low would sit as a fraction of one ATR. Then note which of the three report earnings or have big news due before the next open. Over a few weeks, log every paper stop that triggers and compare the planned price with the price you would actually have gotten. The gap between those two numbers is the risk the stop does not cover.